From 2f443ca7e1ac8e3a08f8494e9a7975b6ffd3c9aa Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 10:28:56 -0700
Subject: [PATCH 01/18] Add nullable Images.Source column for future S3-backed
 image storage

Lays the groundwork for serving images from S3 on demand instead of
requiring a full local sync: an image row can now record where its
bytes live. Existing projects are backfilled via a PRAGMA-guarded
ALTER TABLE in migrateProjectDb, since CREATE TABLE IF NOT EXISTS is
a no-op on tables that already exist.

Co-authored-by: multica-agent <github@multica.ai>
---
 queries/projects/projects.js                | 16 +++++-
 server.js                                   |  7 +++
 tests/integration/imageSourceColumn.test.js | 58 +++++++++++++++++++++
 3 files changed, 80 insertions(+), 1 deletion(-)
 create mode 100644 tests/integration/imageSourceColumn.test.js

diff --git a/queries/projects/projects.js b/queries/projects/projects.js
index 1c5851ef..929b29ee 100644
--- a/queries/projects/projects.js
+++ b/queries/projects/projects.js
@@ -202,7 +202,7 @@ module.exports = {
                 "CREATE TABLE IF NOT EXISTS Classes (CName VARCHAR NOT NULL PRIMARY KEY)",
             );
             await db.run(
-                "CREATE TABLE IF NOT EXISTS Images (IName VARCHAR NOT NULL PRIMARY KEY, reviewImage INTEGER NOT NULL DEFAULT 0, validateImage INTEGER NOT NULL DEFAULT 0)",
+                "CREATE TABLE IF NOT EXISTS Images (IName VARCHAR NOT NULL PRIMARY KEY, reviewImage INTEGER NOT NULL DEFAULT 0, validateImage INTEGER NOT NULL DEFAULT 0, Source VARCHAR DEFAULT NULL)",
             );
             await db.run(
                 "CREATE TABLE IF NOT EXISTS Labels (LID INTEGER PRIMARY KEY, CName VARCHAR NOT NULL, X VARCHAR NOT NULL, Y VARCHAR NOT NULL, W INTEGER NOT NULL, H INTEGER NOT NULL, IName VARCHAR NOT NULL, FOREIGN KEY(CName) REFERENCES Classes(CName), FOREIGN KEY(IName) REFERENCES Images(IName))",
@@ -210,6 +210,20 @@ module.exports = {
             await db.run(
                 "CREATE TABLE IF NOT EXISTS Validation (Confidence INTEGER NOT NULL, LID INTEGER NOT NULL PRIMARY KEY, CName VARCHAR NOT NULL, IName VARCHAR NOT NULL, FOREIGN KEY(LID) REFERENCES Labels(LID), FOREIGN KEY(IName) REFERENCES Images(IName), FOREIGN KEY(CName) REFERENCES Classes(CName))",
             );
+
+            // Images predates the Source column, so CREATE TABLE IF NOT EXISTS above is a
+            // no-op on any project database created before this change. Back-fill it here,
+            // guarded by a PRAGMA check since SQLite has no ADD COLUMN IF NOT EXISTS.
+            const imageColumns = await db.all("PRAGMA table_info(Images)");
+            const hasSourceColumn = (imageColumns.rows || []).some(
+                (column) => column.name === "Source",
+            );
+
+            if (!hasSourceColumn) {
+                await db.run(
+                    "ALTER TABLE Images ADD COLUMN Source VARCHAR DEFAULT NULL",
+                );
+            }
         },
         addImages: async function(
             projectPath,
diff --git a/server.js b/server.js
index ed37eab1..37b0fbfa 100644
--- a/server.js
+++ b/server.js
@@ -1,6 +1,7 @@
 global.logger = require('./utils/logger');
 const app = require('./app');
 const { Client } = require("./queries/client");
+const queries = require("./queries/queries");
 
 global.configFile = require("./utils/config");
 
@@ -49,6 +50,12 @@ for (const project of fs.readdirSync(allProjectsPath)) {
             const dbFile = path.join(projectPath, file);
 
             global.projectDbClients[projectPath] = new Client(dbFile);
+
+            queries.project.migrateProjectDb(projectPath).catch((err) => {
+                global.logger.error(
+                    `Failed to migrate project database at ${projectPath}: ${err}`,
+                );
+            });
         }
     }
 }
diff --git a/tests/integration/imageSourceColumn.test.js b/tests/integration/imageSourceColumn.test.js
new file mode 100644
index 00000000..9e2f4fde
--- /dev/null
+++ b/tests/integration/imageSourceColumn.test.js
@@ -0,0 +1,58 @@
+// Unit tests for queries/projects/projects.js#migrateProjectDb, covering the nullable
+// Images.Source column added so an image's row can record where its bytes live (e.g.
+// local disk vs. an S3 bucket) once storage backends other than local disk are supported.
+// Project databases created before this column existed only get it via the PRAGMA-guarded
+// ALTER TABLE below, since CREATE TABLE IF NOT EXISTS is a no-op on an existing table.
+
+jest.mock('../../queries/getDbClient');
+
+const getDbClient = require('../../queries/getDbClient');
+const projects = require('../../queries/projects/projects');
+
+describe('queries/projects migrateProjectDb Source column backfill', () => {
+    let mockRun;
+    let mockAll;
+
+    beforeEach(() => {
+        jest.clearAllMocks();
+        mockRun = jest.fn().mockResolvedValue({ success: true, changes: 0, lastID: 0 });
+        mockAll = jest.fn();
+        getDbClient.mockReturnValue({ run: mockRun, all: mockAll });
+    });
+
+    it('adds the Source column when an existing Images table predates it', async () => {
+        mockAll.mockResolvedValue({
+            success: true,
+            rows: [
+                { name: 'IName' },
+                { name: 'reviewImage' },
+                { name: 'validateImage' },
+            ],
+        });
+
+        await projects.project.migrateProjectDb('/projects/testuser-test-project');
+
+        expect(mockAll).toHaveBeenCalledWith('PRAGMA table_info(Images)');
+        expect(mockRun).toHaveBeenCalledWith(
+            'ALTER TABLE Images ADD COLUMN Source VARCHAR DEFAULT NULL',
+        );
+    });
+
+    it('does not re-add the Source column when it is already present', async () => {
+        mockAll.mockResolvedValue({
+            success: true,
+            rows: [
+                { name: 'IName' },
+                { name: 'reviewImage' },
+                { name: 'validateImage' },
+                { name: 'Source' },
+            ],
+        });
+
+        await projects.project.migrateProjectDb('/projects/testuser-test-project');
+
+        expect(mockRun).not.toHaveBeenCalledWith(
+            'ALTER TABLE Images ADD COLUMN Source VARCHAR DEFAULT NULL',
+        );
+    });
+});

From 6eafb1a49df08b78d9b29106088c0e3eac722ddc Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 11:42:36 -0700
Subject: [PATCH 02/18] Fix S3 sync silently dropping images that share a
 basename

syncS3Bucket tracked "already synced" purely by sanitized basename, so
two distinct keys under different prefixes (e.g. 2024/img.jpg and
2025/img.jpg) collided and the second was silently skipped.

Adds a nullable Images.SourceKey column (parallel to Source, same
PRAGMA-guarded backfill) to record the literal S3 key an image came
from, decoupled from its display IName. syncS3Bucket now treats
"already synced" as "we've recorded this exact key" rather than "a
file with this name exists," and only disambiguates (via a
deterministic hash of the key) when two keys from the same bucket
listing genuinely collide on basename - a name that already existed
on disk before the run is still skipped as before.

Co-authored-by: multica-agent <github@multica.ai>
---
 queries/projects/projects.js                | 34 +++++---
 routes/api/v2/s3Buckets.js                  | 58 ++++++++++++--
 tests/integration/imageSourceColumn.test.js | 76 ++++++++++++++++--
 tests/integration/s3Buckets.test.js         | 89 +++++++++++++++++++++
 4 files changed, 234 insertions(+), 23 deletions(-)

diff --git a/queries/projects/projects.js b/queries/projects/projects.js
index 929b29ee..8e6f27ec 100644
--- a/queries/projects/projects.js
+++ b/queries/projects/projects.js
@@ -202,7 +202,7 @@ module.exports = {
                 "CREATE TABLE IF NOT EXISTS Classes (CName VARCHAR NOT NULL PRIMARY KEY)",
             );
             await db.run(
-                "CREATE TABLE IF NOT EXISTS Images (IName VARCHAR NOT NULL PRIMARY KEY, reviewImage INTEGER NOT NULL DEFAULT 0, validateImage INTEGER NOT NULL DEFAULT 0, Source VARCHAR DEFAULT NULL)",
+                "CREATE TABLE IF NOT EXISTS Images (IName VARCHAR NOT NULL PRIMARY KEY, reviewImage INTEGER NOT NULL DEFAULT 0, validateImage INTEGER NOT NULL DEFAULT 0, Source VARCHAR DEFAULT NULL, SourceKey VARCHAR DEFAULT NULL)",
             );
             await db.run(
                 "CREATE TABLE IF NOT EXISTS Labels (LID INTEGER PRIMARY KEY, CName VARCHAR NOT NULL, X VARCHAR NOT NULL, Y VARCHAR NOT NULL, W INTEGER NOT NULL, H INTEGER NOT NULL, IName VARCHAR NOT NULL, FOREIGN KEY(CName) REFERENCES Classes(CName), FOREIGN KEY(IName) REFERENCES Images(IName))",
@@ -211,18 +211,26 @@ module.exports = {
                 "CREATE TABLE IF NOT EXISTS Validation (Confidence INTEGER NOT NULL, LID INTEGER NOT NULL PRIMARY KEY, CName VARCHAR NOT NULL, IName VARCHAR NOT NULL, FOREIGN KEY(LID) REFERENCES Labels(LID), FOREIGN KEY(IName) REFERENCES Images(IName), FOREIGN KEY(CName) REFERENCES Classes(CName))",
             );
 
-            // Images predates the Source column, so CREATE TABLE IF NOT EXISTS above is a
-            // no-op on any project database created before this change. Back-fill it here,
-            // guarded by a PRAGMA check since SQLite has no ADD COLUMN IF NOT EXISTS.
+            // Images predates the Source/SourceKey columns, so CREATE TABLE IF NOT EXISTS
+            // above is a no-op on any project database created before this change.
+            // Back-fill them here, guarded by a PRAGMA check since SQLite has no
+            // ADD COLUMN IF NOT EXISTS. SourceKey holds the literal S3 object key (which
+            // may differ from IName once collisions are disambiguated), decoupled from the
+            // display name.
             const imageColumns = await db.all("PRAGMA table_info(Images)");
-            const hasSourceColumn = (imageColumns.rows || []).some(
-                (column) => column.name === "Source",
+            const existingColumnNames = new Set(
+                (imageColumns.rows || []).map((column) => column.name),
             );
 
-            if (!hasSourceColumn) {
-                await db.run(
-                    "ALTER TABLE Images ADD COLUMN Source VARCHAR DEFAULT NULL",
-                );
+            const backfillColumns = [
+                { name: "Source", ddl: "Source VARCHAR DEFAULT NULL" },
+                { name: "SourceKey", ddl: "SourceKey VARCHAR DEFAULT NULL" },
+            ];
+
+            for (const column of backfillColumns) {
+                if (!existingColumnNames.has(column.name)) {
+                    await db.run(`ALTER TABLE Images ADD COLUMN ${column.ddl}`);
+                }
             }
         },
         addImages: async function(
@@ -230,14 +238,18 @@ module.exports = {
             imageName,
             reviewImage,
             validateImage,
+            source = null,
+            sourceKey = null,
         ) {
             const db = getDbClient(projectPath);
             const query =
-                "INSERT OR IGNORE INTO Images (IName, reviewImage, validateImage) VALUES (?, ?, ?)";
+                "INSERT OR IGNORE INTO Images (IName, reviewImage, validateImage, Source, SourceKey) VALUES (?, ?, ?, ?, ?)";
             const results = await db.run(query, [
                 imageName,
                 reviewImage,
                 validateImage,
+                source,
+                sourceKey,
             ]);
 
             return results;
diff --git a/routes/api/v2/s3Buckets.js b/routes/api/v2/s3Buckets.js
index 525fa438..53b49d25 100644
--- a/routes/api/v2/s3Buckets.js
+++ b/routes/api/v2/s3Buckets.js
@@ -1,5 +1,6 @@
 const fs = require("fs");
 const path = require("path");
+const crypto = require("crypto");
 const queries = require("../../../queries/queries");
 const {
     buildS3Client,
@@ -25,6 +26,18 @@ function sanitizeFileName(name) {
         .join("_");
 }
 
+// Two distinct S3 keys can share a basename (e.g. "2024/img.jpg" and
+// "2025/img.jpg"), so a plain basename can't be assumed unique across a
+// bucket. Disambiguate deterministically from the full key so re-running a
+// sync never renames or duplicates a file that was already given a suffix.
+function disambiguateFileName(fileName, key) {
+    const ext = path.extname(fileName);
+    const base = fileName.slice(0, fileName.length - ext.length);
+    const hash = crypto.createHash("sha1").update(key).digest("hex").slice(0, 8);
+
+    return `${base}_${hash}${ext}`;
+}
+
 function trimOrUndefined(value) {
     return typeof value === "string" ? value.trim() : value;
 }
@@ -167,24 +180,59 @@ async function syncS3Bucket(req, res) {
         });
 
         const objectKeys = await listImageObjects(s3Client, bucket.BucketName, bucket.Prefix);
-        const existingImages = new Set(await global.readdirAsync(imagesPath));
+
+        // Snapshot of what's on disk *before* this run - unrelated to what this run
+        // itself assigns. A basename already present here (a prior sync from before
+        // SourceKey existed, or an unrelated local file) is treated as already covered,
+        // same as before. `assignedNames` starts from this snapshot and grows as this
+        // run hands out names, so it also catches two keys *from this same bucket
+        // listing* colliding with each other.
+        const preExistingImages = new Set(await global.readdirAsync(imagesPath));
+        const assignedNames = new Set(preExistingImages);
+
+        const existingImageRows = await queries.project.getAllImages(projectPath);
+        const existingSourceKeys = new Set(
+            (existingImageRows.rows || [])
+                .map((row) => row.SourceKey)
+                .filter(Boolean),
+        );
 
         const syncedImages = [];
         let skippedCount = 0;
 
         for (const key of objectKeys) {
-            const fileName = sanitizeFileName(path.basename(key));
+            // Already synced this exact object in a prior run - never re-download or
+            // re-disambiguate a key we've already assigned a name to.
+            if (existingSourceKeys.has(key)) {
+                skippedCount += 1;
+                continue;
+            }
+
+            const baseName = sanitizeFileName(path.basename(key));
 
-            if (!fileName || existingImages.has(fileName)) {
+            if (!baseName) {
                 skippedCount += 1;
                 continue;
             }
 
+            if (preExistingImages.has(baseName)) {
+                skippedCount += 1;
+                continue;
+            }
+
+            // A different key from this same bucket listing already claimed this
+            // basename - these are genuinely distinct objects, so disambiguate instead
+            // of dropping this one.
+            const fileName = assignedNames.has(baseName)
+                ? disambiguateFileName(baseName, key)
+                : baseName;
+
             const destPath = path.join(imagesPath, fileName);
             await downloadObjectToFile(s3Client, bucket.BucketName, key, destPath);
-            await queries.project.addImages(projectPath, fileName, 0, 0);
+            await queries.project.addImages(projectPath, fileName, 0, 0, "s3", key);
 
-            existingImages.add(fileName);
+            assignedNames.add(fileName);
+            existingSourceKeys.add(key);
             syncedImages.push(fileName);
         }
 
diff --git a/tests/integration/imageSourceColumn.test.js b/tests/integration/imageSourceColumn.test.js
index 9e2f4fde..cd0aebc6 100644
--- a/tests/integration/imageSourceColumn.test.js
+++ b/tests/integration/imageSourceColumn.test.js
@@ -1,15 +1,17 @@
 // Unit tests for queries/projects/projects.js#migrateProjectDb, covering the nullable
-// Images.Source column added so an image's row can record where its bytes live (e.g.
-// local disk vs. an S3 bucket) once storage backends other than local disk are supported.
-// Project databases created before this column existed only get it via the PRAGMA-guarded
-// ALTER TABLE below, since CREATE TABLE IF NOT EXISTS is a no-op on an existing table.
+// Images.Source/SourceKey columns added so an image's row can record where its bytes
+// live (e.g. local disk vs. an S3 bucket) and, for S3-backed images, the literal object
+// key (decoupled from IName, since two different keys can sanitize to the same display
+// name and need disambiguating - see s3Buckets.test.js). Project databases created
+// before these columns existed only get them via the PRAGMA-guarded ALTER TABLE below,
+// since CREATE TABLE IF NOT EXISTS is a no-op on an existing table.
 
 jest.mock('../../queries/getDbClient');
 
 const getDbClient = require('../../queries/getDbClient');
 const projects = require('../../queries/projects/projects');
 
-describe('queries/projects migrateProjectDb Source column backfill', () => {
+describe('queries/projects migrateProjectDb Source/SourceKey column backfill', () => {
     let mockRun;
     let mockAll;
 
@@ -20,7 +22,7 @@ describe('queries/projects migrateProjectDb Source column backfill', () => {
         getDbClient.mockReturnValue({ run: mockRun, all: mockAll });
     });
 
-    it('adds the Source column when an existing Images table predates it', async () => {
+    it('adds both columns when an existing Images table predates them', async () => {
         mockAll.mockResolvedValue({
             success: true,
             rows: [
@@ -36,9 +38,12 @@ describe('queries/projects migrateProjectDb Source column backfill', () => {
         expect(mockRun).toHaveBeenCalledWith(
             'ALTER TABLE Images ADD COLUMN Source VARCHAR DEFAULT NULL',
         );
+        expect(mockRun).toHaveBeenCalledWith(
+            'ALTER TABLE Images ADD COLUMN SourceKey VARCHAR DEFAULT NULL',
+        );
     });
 
-    it('does not re-add the Source column when it is already present', async () => {
+    it('only adds the column that is missing when one already exists', async () => {
         mockAll.mockResolvedValue({
             success: true,
             rows: [
@@ -54,5 +59,62 @@ describe('queries/projects migrateProjectDb Source column backfill', () => {
         expect(mockRun).not.toHaveBeenCalledWith(
             'ALTER TABLE Images ADD COLUMN Source VARCHAR DEFAULT NULL',
         );
+        expect(mockRun).toHaveBeenCalledWith(
+            'ALTER TABLE Images ADD COLUMN SourceKey VARCHAR DEFAULT NULL',
+        );
+    });
+
+    it('does not re-add either column when both are already present', async () => {
+        mockAll.mockResolvedValue({
+            success: true,
+            rows: [
+                { name: 'IName' },
+                { name: 'reviewImage' },
+                { name: 'validateImage' },
+                { name: 'Source' },
+                { name: 'SourceKey' },
+            ],
+        });
+
+        await projects.project.migrateProjectDb('/projects/testuser-test-project');
+
+        expect(mockRun).not.toHaveBeenCalledWith(
+            expect.stringContaining('ALTER TABLE Images ADD COLUMN'),
+        );
+    });
+});
+
+describe('queries/projects addImages Source/SourceKey parameters', () => {
+    let mockRun;
+
+    beforeEach(() => {
+        jest.clearAllMocks();
+        mockRun = jest.fn().mockResolvedValue({ success: true, changes: 1, lastID: 1 });
+        getDbClient.mockReturnValue({ run: mockRun });
+    });
+
+    it('defaults Source and SourceKey to null for callers that do not pass them', async () => {
+        await projects.project.addImages('/projects/testuser-test-project', 'cat.jpg', 0, 0);
+
+        expect(mockRun).toHaveBeenCalledWith(
+            expect.stringContaining('INSERT OR IGNORE INTO Images'),
+            ['cat.jpg', 0, 0, null, null],
+        );
+    });
+
+    it('passes through Source and SourceKey when a caller (e.g. S3 sync) provides them', async () => {
+        await projects.project.addImages(
+            '/projects/testuser-test-project',
+            'cat.jpg',
+            0,
+            0,
+            's3',
+            'images/cat.jpg',
+        );
+
+        expect(mockRun).toHaveBeenCalledWith(
+            expect.stringContaining('INSERT OR IGNORE INTO Images'),
+            ['cat.jpg', 0, 0, 's3', 'images/cat.jpg'],
+        );
     });
 });
diff --git a/tests/integration/s3Buckets.test.js b/tests/integration/s3Buckets.test.js
index 94d7caab..fc4bb3b2 100644
--- a/tests/integration/s3Buckets.test.js
+++ b/tests/integration/s3Buckets.test.js
@@ -11,6 +11,7 @@ jest.mock('../../queries/queries', () => ({
     },
     project: {
         addImages: jest.fn().mockResolvedValue({ success: true }),
+        getAllImages: jest.fn().mockResolvedValue({ success: true, rows: [] }),
     },
 }));
 
@@ -231,10 +232,98 @@ describe('S3 Bucket Routes', () => {
                 'cat.jpg',
                 0,
                 0,
+                's3',
+                'images/cat.jpg',
             );
             expect(queries.managed.touchBucketSyncedAt).toHaveBeenCalled();
         });
 
+        it('skips an object already synced by key, even if its name is no longer the only file on disk', async () => {
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: {
+                    BucketName: 'my-bucket',
+                    Region: 'us-east-1',
+                    Prefix: 'images/',
+                    AccessKeyId: 'AKIA...',
+                    SecretAccessKey: 'secret',
+                },
+            });
+            s3Client.listImageObjects.mockResolvedValueOnce(['images/cat.jpg']);
+            global.readdirAsync.mockResolvedValueOnce(['cat.jpg']);
+            queries.project.getAllImages.mockResolvedValueOnce({
+                success: true,
+                rows: [{ IName: 'cat.jpg', Source: 's3', SourceKey: 'images/cat.jpg' }],
+            });
+
+            const res = await request(app)
+                .post('/api/v2/projects/testuser/test-project/s3-bucket/sync')
+                .set('Cookie', ['Username=testuser']);
+
+            expect(res.statusCode).toBe(200);
+            expect(res.body.syncedCount).toBe(0);
+            expect(res.body.skippedCount).toBe(1);
+            expect(s3Client.downloadObjectToFile).not.toHaveBeenCalled();
+            expect(queries.project.addImages).not.toHaveBeenCalled();
+        });
+
+        it('disambiguates two different keys that sanitize to the same basename instead of dropping one', async () => {
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: {
+                    BucketName: 'my-bucket',
+                    Region: 'us-east-1',
+                    Prefix: 'images/',
+                    AccessKeyId: 'AKIA...',
+                    SecretAccessKey: 'secret',
+                },
+            });
+            s3Client.listImageObjects.mockResolvedValueOnce([
+                '2024/img.jpg',
+                '2025/img.jpg',
+            ]);
+            global.readdirAsync.mockResolvedValueOnce([]);
+            queries.project.getAllImages.mockResolvedValueOnce({ success: true, rows: [] });
+
+            const res = await request(app)
+                .post('/api/v2/projects/testuser/test-project/s3-bucket/sync')
+                .set('Cookie', ['Username=testuser']);
+
+            expect(res.statusCode).toBe(200);
+            expect(res.body.syncedCount).toBe(2);
+            expect(res.body.skippedCount).toBe(0);
+            expect(res.body.images[0]).toBe('img.jpg');
+            expect(res.body.images[1]).not.toBe('img.jpg');
+            expect(res.body.images[1]).toMatch(/^img_[0-9a-f]{8}\.jpg$/);
+
+            expect(queries.project.addImages).toHaveBeenNthCalledWith(
+                1,
+                expect.stringContaining('testuser-test-project'),
+                'img.jpg',
+                0,
+                0,
+                's3',
+                '2024/img.jpg',
+            );
+            expect(queries.project.addImages).toHaveBeenNthCalledWith(
+                2,
+                expect.stringContaining('testuser-test-project'),
+                res.body.images[1],
+                0,
+                0,
+                's3',
+                '2025/img.jpg',
+            );
+
+            // Deterministic: re-running sync against the same key must reproduce the
+            // same disambiguated name, not draw a fresh suffix each time.
+            const crypto = require('crypto');
+            const expectedHash = crypto
+                .createHash('sha1')
+                .update('2025/img.jpg')
+                .digest('hex')
+                .slice(0, 8);
+            expect(res.body.images[1]).toBe(`img_${expectedHash}.jpg`);
+        });
+
         it('returns 404 when the project has no attached bucket', async () => {
             queries.managed.getBucket.mockResolvedValueOnce({ row: undefined });
 

From a05d6ccb84a3054974e996ccf4aa53df10f7511c Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 11:49:54 -0700
Subject: [PATCH 03/18] Downgrade SQLITE_READONLY to a warning during startup
 project migration

migrateProjectDb now runs for every existing project at server boot
(added alongside the Source/SourceKey columns), and every statement
in it is DDL - even a no-op CREATE TABLE IF NOT EXISTS still requires
SQLite to open the file for write. A read-only project database can
never be migrated regardless of the reason it's read-only, so treat
that specific, expected outcome as a warning instead of an
error-level log repeated on every single server start.

Co-authored-by: multica-agent <github@multica.ai>
---
 server.js | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/server.js b/server.js
index 37b0fbfa..0b213fb3 100644
--- a/server.js
+++ b/server.js
@@ -52,6 +52,20 @@ for (const project of fs.readdirSync(allProjectsPath)) {
             global.projectDbClients[projectPath] = new Client(dbFile);
 
             queries.project.migrateProjectDb(projectPath).catch((err) => {
+                // Every statement in migrateProjectDb is DDL (even a no-op
+                // CREATE TABLE IF NOT EXISTS), so SQLite opens the file for
+                // write regardless of whether a change is actually needed.
+                // A read-only project database - by design, or a permissions
+                // quirk of wherever it's deployed - can't be migrated, but
+                // that's expected and not a failure worth an error-level log
+                // on every server start.
+                if (err && err.error && err.error.code === "SQLITE_READONLY") {
+                    global.logger.warn(
+                        `Project database at ${projectPath} is read-only; skipping schema migration.`,
+                    );
+                    return;
+                }
+
                 global.logger.error(
                     `Failed to migrate project database at ${projectPath}: ${err}`,
                 );

From 28be9501ca4483d3593bf26073c7c7ac2ae0474f Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 12:57:56 -0700
Subject: [PATCH 04/18] Map container UID/GID to the host user in
 docker-compose

Root-caused the SQLITE_READONLY errors on startup migration: the app
image has no USER directive, so the container writes to the
bind-mounted public/projects (and db, data) directories as root.
Anything created while running via docker-compose then can't be
written by a normal host user running the app outside Docker
(`node .`), which is exactly what surfaced as SQLITE_READONLY for
test-gcp-test/test-gcp-test3. Mapping the container to the host
UID/GID prevents this going forward; already-affected directories
still need a one-time chown.

Co-authored-by: multica-agent <github@multica.ai>
---
 docker-compose.yml | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/docker-compose.yml b/docker-compose.yml
index b32c102f..220ce1ff 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -6,6 +6,14 @@ services:
     image: ghcr.io/${GITHUB_REPOSITORY:-njobvu-ai}/app:latest
     container_name: njobvu-ai
     restart: unless-stopped
+    # The image has no USER directive, so without this the container writes
+    # to the bind-mounted host directories below as root. Anything it
+    # creates (e.g. project databases/images) then can't be written by a
+    # normal host user running the app outside Docker (`node .`), failing
+    # with SQLITE_READONLY. Export UID/GID before `docker compose up`
+    # (most shells: `export UID GID` - `UID` is a shell built-in that isn't
+    # exported by default) so files it creates are owned by you instead.
+    user: "${UID:-1000}:${GID:-1000}"
     ports:
       - "3000:3000"
     environment:

From d5f80b410709f93769527f004280a16b0dca070b Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 12:58:08 -0700
Subject: [PATCH 05/18] Add a download-vs-stream choice for S3 bucket sync

Adds a SyncMode ("download" | "stream") on S3Buckets, chosen via a new
checkbox in the S3 project-creation form. "download" keeps today's
behavior unchanged. "stream" registers each object as a project image
without copying it to local disk, and adds an on-demand image proxy
(GET /api/v2/projects/:admin/:projectName/images/:imageName) that
serves a local file if present, or otherwise fetches the object live
from S3 for that one request - never persisting it.

getAnnotatePage (the per-image labeling/annotation view) now falls
back to that proxy and probes the object's dimensions over the same
live fetch (aborted early once the header is read, via
probe-image-size's stream support) instead of 404ing when an image
was never downloaded.

Co-authored-by: multica-agent <github@multica.ai>
---
 db/migrations.sql                             |   1 +
 queries/s3/s3.js                              |  11 +-
 routes/api.js                                 |   2 +
 routes/api/v2/s3Buckets.js                    |  82 ++++++++-
 routes/pages/getAnnotatePage.js               |  70 +++++---
 tests/integration/s3Buckets.test.js           | 149 +++++++++++++++++
 tests/unit/getAnnotatePageS3Streaming.test.js | 155 ++++++++++++++++++
 utils/s3Client.js                             |  16 ++
 views/create.ejs                              |  13 +-
 9 files changed, 472 insertions(+), 27 deletions(-)
 create mode 100644 tests/unit/getAnnotatePageS3Streaming.test.js

diff --git a/db/migrations.sql b/db/migrations.sql
index 864395ac..8a0431a5 100644
--- a/db/migrations.sql
+++ b/db/migrations.sql
@@ -3,3 +3,4 @@ INSERT INTO Users (Username, Password, FirstName, LastName, Email) VALUES ('Zero
 CREATE TABLE Projects (PName VARCHAR NOT NULL, PDescription VARCHAR NOT NULL, AutoSave INTEGER NOT NULL DEFAULT 0, Admin TEXT NOT NULL DEFAULT 'ZeroUser', Validate VARCHAR NOT NULL DEFAULT 0, FOREIGN KEY(Admin) REFERENCES Users(Username), PRIMARY KEY(PName, Admin));
 CREATE TABLE Access (Username TEXT NOT NULL, PName VARCHAR NOT NULL,  Admin TEXT NOT NULL, FOREIGN KEY(Username) REFERENCES Users(Username), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
 CREATE TABLE S3Buckets (PName VARCHAR NOT NULL, Admin TEXT NOT NULL, BucketName VARCHAR NOT NULL, Region VARCHAR NOT NULL, Prefix VARCHAR NOT NULL DEFAULT '', AccessKeyId VARCHAR, SecretAccessKey VARCHAR, LastSyncedAt TEXT, Endpoint TEXT, PRIMARY KEY(PName, Admin), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
+ALTER TABLE S3Buckets ADD COLUMN SyncMode VARCHAR NOT NULL DEFAULT 'download';
diff --git a/queries/s3/s3.js b/queries/s3/s3.js
index 5c5fe857..88d0e182 100644
--- a/queries/s3/s3.js
+++ b/queries/s3/s3.js
@@ -8,14 +8,16 @@ module.exports = {
             prefix,
             accessKeyId,
             secretAccessKey,
-            endpoint
+            endpoint,
+            syncMode = "download",
         ) {
             const query =
-                "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint) " +
-                "VALUES (?, ?, ?, ?, ?, ?, ?, ?) " +
+                "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint, SyncMode) " +
+                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?) " +
                 "ON CONFLICT(PName, Admin) DO UPDATE SET " +
                 "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
-                "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint";
+                "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint, " +
+                "SyncMode = excluded.SyncMode";
 
             const result = await global.managedDbClient.run(query, [
                 projectName,
@@ -26,6 +28,7 @@ module.exports = {
                 accessKeyId || null,
                 secretAccessKey || null,
                 endpoint || "",
+                syncMode,
             ]);
 
             return result;
diff --git a/routes/api.js b/routes/api.js
index 72f9bd42..31c3a221 100755
--- a/routes/api.js
+++ b/routes/api.js
@@ -71,6 +71,7 @@ const {
     getS3Bucket,
     deleteS3Bucket,
     syncS3Bucket,
+    getProjectImage,
 } = require("./api/v2/s3Buckets");
 
 const updateLabels = require("./labelling/updateLabels");
@@ -188,6 +189,7 @@ api.post("/api/v2/projects/:admin/:projectName/s3-bucket", attachS3Bucket);
 api.get("/api/v2/projects/:admin/:projectName/s3-bucket", getS3Bucket);
 api.delete("/api/v2/projects/:admin/:projectName/s3-bucket", deleteS3Bucket);
 api.post("/api/v2/projects/:admin/:projectName/s3-bucket/sync", syncS3Bucket);
+api.get("/api/v2/projects/:admin/:projectName/images/:imageName", getProjectImage);
 
 // LABELLING ROUTES
 api.post("/updateLabels", updateLabels);
diff --git a/routes/api/v2/s3Buckets.js b/routes/api/v2/s3Buckets.js
index 53b49d25..bc724f70 100644
--- a/routes/api/v2/s3Buckets.js
+++ b/routes/api/v2/s3Buckets.js
@@ -7,8 +7,11 @@ const {
     verifyBucketAccess,
     listImageObjects,
     downloadObjectToFile,
+    getObjectStream,
 } = require("../../../utils/s3Client");
 
+const SYNC_MODES = new Set(["download", "stream"]);
+
 function getProjectPath(admin, projectName) {
     return path.join(currentPath, "public", "projects", `${admin}-${projectName}`);
 }
@@ -51,6 +54,7 @@ async function attachS3Bucket(req, res) {
     const AccessKeyId = trimOrUndefined(body.AccessKeyId);
     const SecretAccessKey = trimOrUndefined(body.SecretAccessKey);
     const Endpoint = trimOrUndefined(body.Endpoint);
+    const SyncMode = SYNC_MODES.has(body.SyncMode) ? body.SyncMode : "download";
 
     if (!isOwner(req, admin)) {
         return res.status(403).json({ success: false, error: "Not authorized for this project" });
@@ -83,7 +87,8 @@ async function attachS3Bucket(req, res) {
             Prefix || "",
             AccessKeyId,
             SecretAccessKey,
-            ...(Endpoint ? [Endpoint] : []),
+            Endpoint || "",
+            SyncMode,
         );
 
         return res.status(200).json({ success: true });
@@ -123,6 +128,7 @@ async function getS3Bucket(req, res) {
                 Prefix: row.Prefix,
                 Endpoint: row.Endpoint,
                 LastSyncedAt: row.LastSyncedAt,
+                SyncMode: row.SyncMode || "download",
                 hasCredentials: !!row.AccessKeyId,
             },
         });
@@ -227,8 +233,14 @@ async function syncS3Bucket(req, res) {
                 ? disambiguateFileName(baseName, key)
                 : baseName;
 
-            const destPath = path.join(imagesPath, fileName);
-            await downloadObjectToFile(s3Client, bucket.BucketName, key, destPath);
+            // "stream" mode registers the image (so it's browsable/labelable) without
+            // pulling its bytes to disk - GET .../images/:imageName fetches them from
+            // the bucket live, on the rare request that actually needs them.
+            if (bucket.SyncMode !== "stream") {
+                const destPath = path.join(imagesPath, fileName);
+                await downloadObjectToFile(s3Client, bucket.BucketName, key, destPath);
+            }
+
             await queries.project.addImages(projectPath, fileName, 0, 0, "s3", key);
 
             assignedNames.add(fileName);
@@ -254,9 +266,73 @@ async function syncS3Bucket(req, res) {
     }
 }
 
+// Serves a single project image, transparently covering both storage modes:
+// a locally-present file (the common case - local imports, or "download"-mode
+// S3 sync) is sent straight from disk; an image registered from S3 that was
+// never downloaded ("stream" mode) is fetched from the bucket live, for this
+// request only, and never written to disk. Any logged-in user can view it -
+// this mirrors the existing (also unauthenticated) static file serving for
+// local images, rather than introducing a stricter, inconsistent check here.
+async function getProjectImage(req, res) {
+    const { admin, projectName, imageName } = req.params;
+
+    if (!req.cookies || !req.cookies.Username) {
+        return res.status(403).json({ success: false, error: "Not authorized" });
+    }
+
+    const projectPath = getProjectPath(admin, projectName);
+    const localPath = path.join(projectPath, "images", imageName);
+
+    if (fs.existsSync(localPath)) {
+        return res.sendFile(localPath);
+    }
+
+    try {
+        const imageResult = await queries.project.getImage(projectPath, imageName);
+        const image = imageResult && imageResult.row;
+
+        if (!image || image.Source !== "s3" || !image.SourceKey) {
+            return res.status(404).json({ success: false, error: "Image not found" });
+        }
+
+        const bucketResult = await queries.managed.getBucket(projectName, admin);
+        const bucket = bucketResult && bucketResult.row;
+
+        if (!bucket) {
+            return res.status(404).json({ success: false, error: "No S3 bucket attached to this project" });
+        }
+
+        const s3Client = buildS3Client({
+            region: bucket.Region,
+            accessKeyId: bucket.AccessKeyId,
+            secretAccessKey: bucket.SecretAccessKey,
+            endpoint: bucket.Endpoint,
+        });
+
+        const { body, contentType } = await getObjectStream(s3Client, bucket.BucketName, image.SourceKey);
+
+        res.setHeader("Content-Type", contentType || "application/octet-stream");
+        body.on("error", (err) => {
+            global.logger.error(err);
+            if (!res.headersSent) {
+                res.status(500).end();
+            }
+        });
+        body.pipe(res);
+    } catch (err) {
+        global.logger.error(err, {
+            httpStatusCode: err.$metadata?.httpStatusCode,
+            code: err.Code || err.name,
+            requestId: err.$metadata?.requestId,
+        });
+        return res.status(500).json({ success: false, error: "Error fetching image" });
+    }
+}
+
 module.exports = {
     attachS3Bucket,
     getS3Bucket,
     deleteS3Bucket,
     syncS3Bucket,
+    getProjectImage,
 };
diff --git a/routes/pages/getAnnotatePage.js b/routes/pages/getAnnotatePage.js
index 0bd35b6f..f043e512 100644
--- a/routes/pages/getAnnotatePage.js
+++ b/routes/pages/getAnnotatePage.js
@@ -1,3 +1,6 @@
+const queries = require("../../queries/queries");
+const { buildS3Client, getObjectStream } = require("../../utils/s3Client");
+
 async function getAnnotatePage(req, res) {
     var IDX = parseInt(req.query.IDX),
         IName = String(req.query.IName),
@@ -129,30 +132,59 @@ async function getAnnotatePage(req, res) {
     }
 
     var abs_image_path = project_path + "/images/" + IName;
-
-    if (!results4 || results4.length === 0 || !fs.existsSync(abs_image_path)) {
+    var imageExistsLocally = fs.existsSync(abs_image_path);
+    var imageRow = results4 && results4[0];
+
+    // A "download"-mode (or local-import) image is a real file at
+    // abs_image_path, same as always. A "stream"-mode S3 image never has
+    // one - it's only ever fetched live, on view, via the on-demand proxy
+    // below - so only 404 here if neither a local file nor an S3-backed row
+    // exists for this name.
+    if (!imageRow || (!imageExistsLocally && imageRow.Source !== "s3")) {
         res.render("404", {
             title: "404",
             user: req.cookies.Username,
         });
     } else {
-        var rel_image_path = rel_project_path + "/images/" + results4[0].IName;
-        var img = fs.readFileSync(
-                project_path + "/images/" + results4[0].IName,
-                (err) => {
-                    if (err) {
-                        res.render("404", {
-                            title: "404",
-                            user: req.cookies.Username,
-                        });
-                    }
-                },
-            ),
-            img_data = probe.sync(img),
-            img_w = img_data.width,
-            img_h = img_data.height,
-            image_ratio = img_h / img_w,
-            image_width = img_w,
+        var rel_image_path;
+        var img_w, img_h;
+
+        if (imageExistsLocally) {
+            rel_image_path = rel_project_path + "/images/" + imageRow.IName;
+            var img = fs.readFileSync(project_path + "/images/" + imageRow.IName);
+            var img_data = probe.sync(img);
+            img_w = img_data.width;
+            img_h = img_data.height;
+        } else {
+            // Point the browser at the on-demand proxy instead of a static
+            // path that doesn't exist. Probe just enough of the object's
+            // header (over that same live fetch, aborted by probe() once it
+            // has what it needs) to lay out the page, rather than pulling
+            // the whole image server-side just to measure it.
+            rel_image_path = `api/v2/projects/${admin}/${PName}/images/${imageRow.IName}`;
+
+            try {
+                var bucketResult = await queries.managed.getBucket(PName, admin);
+                var bucket = bucketResult && bucketResult.row;
+                var s3Client = buildS3Client({
+                    region: bucket.Region,
+                    accessKeyId: bucket.AccessKeyId,
+                    secretAccessKey: bucket.SecretAccessKey,
+                    endpoint: bucket.Endpoint,
+                });
+                var objectStream = await getObjectStream(s3Client, bucket.BucketName, imageRow.SourceKey);
+                var probed = await probe(objectStream.body);
+                img_w = probed.width;
+                img_h = probed.height;
+            } catch (err) {
+                global.logger.error(err);
+                img_w = 0;
+                img_h = 0;
+            }
+        }
+
+        var image_ratio = img_w ? img_h / img_w : 1,
+            image_width = img_w || 0,
             image_height = image_ratio * image_width,
             prev_IName = (next_IName = -1);
         var curr_index = 1;
diff --git a/tests/integration/s3Buckets.test.js b/tests/integration/s3Buckets.test.js
index fc4bb3b2..6ca28e93 100644
--- a/tests/integration/s3Buckets.test.js
+++ b/tests/integration/s3Buckets.test.js
@@ -12,6 +12,7 @@ jest.mock('../../queries/queries', () => ({
     project: {
         addImages: jest.fn().mockResolvedValue({ success: true }),
         getAllImages: jest.fn().mockResolvedValue({ success: true, rows: [] }),
+        getImage: jest.fn(),
     },
 }));
 
@@ -20,8 +21,10 @@ jest.mock('../../utils/s3Client', () => ({
     verifyBucketAccess: jest.fn().mockResolvedValue(undefined),
     listImageObjects: jest.fn().mockResolvedValue([]),
     downloadObjectToFile: jest.fn().mockResolvedValue(undefined),
+    getObjectStream: jest.fn(),
 }));
 
+const { Readable } = require('stream');
 const request = require('supertest');
 const express = require('express');
 const cookieParser = require('cookie-parser');
@@ -33,6 +36,7 @@ const {
     getS3Bucket,
     deleteS3Bucket,
     syncS3Bucket,
+    getProjectImage,
 } = require('../../routes/api/v2/s3Buckets');
 
 // These handlers are exercised directly against a minimal app (rather than
@@ -49,6 +53,17 @@ function buildTestApp() {
     app.get('/api/v2/projects/:admin/:projectName/s3-bucket', getS3Bucket);
     app.delete('/api/v2/projects/:admin/:projectName/s3-bucket', deleteS3Bucket);
     app.post('/api/v2/projects/:admin/:projectName/s3-bucket/sync', syncS3Bucket);
+    // `fs` is mocked module-wide down to just existsSync (below), so stub
+    // res.sendFile here rather than letting Express's real implementation
+    // reach for fs.stat/createReadStream, which don't exist on the mock.
+    app.get(
+        '/api/v2/projects/:admin/:projectName/images/:imageName',
+        (req, res, next) => {
+            res.sendFile = jest.fn((filePath) => res.status(200).send(`local-file:${filePath}`));
+            next();
+        },
+        getProjectImage,
+    );
 
     return app;
 }
@@ -97,6 +112,30 @@ describe('S3 Bucket Routes', () => {
                 'images/',
                 'AKIA...',
                 'secret',
+                '',
+                'download',
+            );
+        });
+
+        it('defaults SyncMode to "download" when omitted or invalid, and passes through "stream" when requested', async () => {
+            await request(app)
+                .post('/api/v2/projects/testuser/test-project/s3-bucket')
+                .set('Cookie', ['Username=testuser'])
+                .send({ ...validBody, SyncMode: 'not-a-real-mode' });
+
+            expect(queries.managed.attachBucket).toHaveBeenNthCalledWith(
+                1,
+                'test-project', 'testuser', 'my-bucket', 'us-east-1', 'images/', 'AKIA...', 'secret', '', 'download',
+            );
+
+            await request(app)
+                .post('/api/v2/projects/testuser/test-project/s3-bucket')
+                .set('Cookie', ['Username=testuser'])
+                .send({ ...validBody, SyncMode: 'stream' });
+
+            expect(queries.managed.attachBucket).toHaveBeenNthCalledWith(
+                2,
+                'test-project', 'testuser', 'my-bucket', 'us-east-1', 'images/', 'AKIA...', 'secret', '', 'stream',
             );
         });
 
@@ -155,6 +194,7 @@ describe('S3 Bucket Routes', () => {
                     LastSyncedAt: null,
                     AccessKeyId: 'AKIA...',
                     SecretAccessKey: 'super-secret',
+                    SyncMode: 'stream',
                 },
             });
 
@@ -164,11 +204,24 @@ describe('S3 Bucket Routes', () => {
 
             expect(res.statusCode).toBe(200);
             expect(res.body.bucket.BucketName).toBe('my-bucket');
+            expect(res.body.bucket.SyncMode).toBe('stream');
             expect(res.body.bucket.hasCredentials).toBe(true);
             expect(res.body.bucket.SecretAccessKey).toBeUndefined();
             expect(res.body.bucket.AccessKeyId).toBeUndefined();
         });
 
+        it('defaults SyncMode to "download" for buckets attached before the column existed', async () => {
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: { BucketName: 'my-bucket', Region: 'us-east-1', Prefix: '', SyncMode: null },
+            });
+
+            const res = await request(app)
+                .get('/api/v2/projects/testuser/test-project/s3-bucket')
+                .set('Cookie', ['Username=testuser']);
+
+            expect(res.body.bucket.SyncMode).toBe('download');
+        });
+
         it('returns 404 when no bucket is attached', async () => {
             queries.managed.getBucket.mockResolvedValueOnce({ row: undefined });
 
@@ -345,5 +398,101 @@ describe('S3 Bucket Routes', () => {
             expect(res.statusCode).toBe(404);
             expect(queries.managed.getBucket).not.toHaveBeenCalled();
         });
+
+        it('registers images without downloading them when SyncMode is "stream"', async () => {
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: {
+                    BucketName: 'my-bucket',
+                    Region: 'us-east-1',
+                    Prefix: 'images/',
+                    AccessKeyId: 'AKIA...',
+                    SecretAccessKey: 'secret',
+                    SyncMode: 'stream',
+                },
+            });
+            s3Client.listImageObjects.mockResolvedValueOnce(['images/cat.jpg']);
+            global.readdirAsync.mockResolvedValueOnce([]);
+
+            const res = await request(app)
+                .post('/api/v2/projects/testuser/test-project/s3-bucket/sync')
+                .set('Cookie', ['Username=testuser']);
+
+            expect(res.statusCode).toBe(200);
+            expect(res.body.syncedCount).toBe(1);
+            expect(s3Client.downloadObjectToFile).not.toHaveBeenCalled();
+            expect(queries.project.addImages).toHaveBeenCalledWith(
+                expect.stringContaining('testuser-test-project'),
+                'cat.jpg',
+                0,
+                0,
+                's3',
+                'images/cat.jpg',
+            );
+        });
+    });
+
+    describe('GET /api/v2/projects/:admin/:projectName/images/:imageName', () => {
+        it('serves a locally-present file from disk without touching S3', async () => {
+            const res = await request(app)
+                .get('/api/v2/projects/testuser/test-project/images/cat.jpg')
+                .set('Cookie', ['Username=testuser']);
+
+            expect(res.statusCode).toBe(200);
+            expect(res.text).toContain('local-file:');
+            expect(res.text).toContain('cat.jpg');
+            expect(s3Client.getObjectStream).not.toHaveBeenCalled();
+        });
+
+        it('streams a "stream"-mode image live from S3 when no local file exists', async () => {
+            fs.existsSync.mockReturnValue(false);
+            queries.project.getImage.mockResolvedValueOnce({
+                row: { IName: 'cat.jpg', Source: 's3', SourceKey: 'images/cat.jpg' },
+            });
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: {
+                    BucketName: 'my-bucket',
+                    Region: 'us-east-1',
+                    AccessKeyId: 'AKIA...',
+                    SecretAccessKey: 'secret',
+                    SyncMode: 'stream',
+                },
+            });
+            s3Client.getObjectStream.mockResolvedValueOnce({
+                body: Readable.from([Buffer.from('fake-image-bytes')]),
+                contentType: 'image/jpeg',
+            });
+
+            const res = await request(app)
+                .get('/api/v2/projects/testuser/test-project/images/cat.jpg')
+                .set('Cookie', ['Username=testuser']);
+
+            expect(res.statusCode).toBe(200);
+            expect(res.headers['content-type']).toBe('image/jpeg');
+            expect(Buffer.from(res.body).toString()).toBe('fake-image-bytes');
+            expect(s3Client.getObjectStream).toHaveBeenCalledWith(
+                { fakeClient: true },
+                'my-bucket',
+                'images/cat.jpg',
+            );
+        });
+
+        it('returns 404 when neither a local file nor an S3-backed row exists', async () => {
+            fs.existsSync.mockReturnValue(false);
+            queries.project.getImage.mockResolvedValueOnce({ row: undefined });
+
+            const res = await request(app)
+                .get('/api/v2/projects/testuser/test-project/images/missing.jpg')
+                .set('Cookie', ['Username=testuser']);
+
+            expect(res.statusCode).toBe(404);
+        });
+
+        it('rejects requests from a logged-out client', async () => {
+            const res = await request(app).get(
+                '/api/v2/projects/testuser/test-project/images/cat.jpg',
+            );
+
+            expect(res.statusCode).toBe(403);
+        });
     });
 });
diff --git a/tests/unit/getAnnotatePageS3Streaming.test.js b/tests/unit/getAnnotatePageS3Streaming.test.js
new file mode 100644
index 00000000..a1f3931f
--- /dev/null
+++ b/tests/unit/getAnnotatePageS3Streaming.test.js
@@ -0,0 +1,155 @@
+// Unit tests for routes/pages/getAnnotatePage.js's S3 "stream" mode handling.
+//
+// Calls the handler directly (bypassing the full Express app/static
+// middleware) since that stack is heavily mocked elsewhere for unrelated
+// tests in a way that intercepts every route with an empty static response -
+// exercising the real branching logic here needs a narrower harness.
+
+jest.mock('../../queries/queries', () => ({
+    managed: {
+        getBucket: jest.fn(),
+    },
+}));
+
+jest.mock('../../utils/s3Client', () => ({
+    buildS3Client: jest.fn(() => ({ fakeClient: true })),
+    getObjectStream: jest.fn(),
+}));
+
+const { Readable } = require('stream');
+const queries = require('../../queries/queries');
+const s3Client = require('../../utils/s3Client');
+const getAnnotatePage = require('../../routes/pages/getAnnotatePage');
+
+// Mirrors the real sqlite3 driver's callback style, since getAnnotatePage.js
+// wraps `this.get`/`this.all` in its own Promise-returning getAsync/allAsync
+// right after construction - only the raw callback methods are ever called.
+function makeFakeProjectDb({ classesRows = [], labelsRows = [], imagesRows = [], displayRow }) {
+    return {
+        get: jest.fn((sql, cb) => {
+            if (sql.includes('display_id')) return cb(null, displayRow);
+            return cb(null, undefined);
+        }),
+        all: jest.fn((sql, cb) => {
+            if (sql.includes('Classes')) return cb(null, classesRows);
+            if (sql.includes('Labels')) return cb(null, labelsRows);
+            if (sql.includes('Images')) return cb(null, imagesRows);
+            return cb(null, []);
+        }),
+        each: jest.fn((sql, cb) => cb(null, undefined)),
+        close: jest.fn((cb) => cb && cb(null)),
+    };
+}
+
+describe('getAnnotatePage - S3-backed image serving', () => {
+    let res;
+    let req;
+
+    beforeEach(() => {
+        jest.clearAllMocks();
+
+        global.logger = { debug: jest.fn(), error: jest.fn(), info: jest.fn() };
+        global.currentPath = '/app/';
+        global.colorsJSON = [{ value: '#FF0000' }];
+        global.db = {
+            allAsync: jest.fn().mockResolvedValue([{ PName: 'test-project', Admin: 'testuser' }]),
+            getAsync: jest.fn().mockResolvedValue({ AutoSave: 1 }),
+        };
+
+        req = {
+            query: { IDX: '0', IName: 'image1.jpg', curr_class: 'class1' },
+            cookies: { Username: 'testuser' },
+        };
+        res = { redirect: jest.fn(), render: jest.fn() };
+    });
+
+    it('serves a "stream"-mode image with no local file via the on-demand S3 proxy, without touching disk', async () => {
+        const imageRow = { IName: 'image1.jpg', reviewImage: 0, Source: 's3', SourceKey: 'images/image1.jpg' };
+
+        global.fs = { existsSync: jest.fn().mockReturnValue(false), readFileSync: jest.fn() };
+        global.sqlite3 = {
+            Database: jest.fn((dbPath, cb) => {
+                cb && cb(null);
+                return makeFakeProjectDb({
+                    imagesRows: [imageRow],
+                    displayRow: { IName: 'image1.jpg', display_id: 1 },
+                });
+            }),
+        };
+        global.probe = jest.fn().mockResolvedValue({ width: 400, height: 300 });
+        global.probe.sync = jest.fn();
+
+        queries.managed.getBucket.mockResolvedValueOnce({
+            row: { BucketName: 'my-bucket', Region: 'us-east-1', AccessKeyId: 'AKIA...', SecretAccessKey: 'secret' },
+        });
+        s3Client.getObjectStream.mockResolvedValueOnce({
+            body: Readable.from([Buffer.from('fake-image-bytes')]),
+            contentType: 'image/jpeg',
+        });
+
+        await getAnnotatePage(req, res);
+
+        expect(global.fs.readFileSync).not.toHaveBeenCalled();
+        expect(global.probe.sync).not.toHaveBeenCalled();
+        expect(s3Client.getObjectStream).toHaveBeenCalledWith(
+            { fakeClient: true },
+            'my-bucket',
+            'images/image1.jpg',
+        );
+        expect(res.render).toHaveBeenCalledWith('annotate', expect.objectContaining({
+            image_path: 'api/v2/projects/testuser/test-project/images/image1.jpg',
+            image_width: 400,
+            image_height: 300,
+        }));
+    });
+
+    it('still reads a locally-present file straight from disk, unaffected by the S3 changes', async () => {
+        const imageRow = { IName: 'image1.jpg', reviewImage: 0, Source: null, SourceKey: null };
+
+        global.fs = {
+            existsSync: jest.fn().mockReturnValue(true),
+            readFileSync: jest.fn().mockReturnValue(Buffer.from('img-bytes')),
+        };
+        global.sqlite3 = {
+            Database: jest.fn((dbPath, cb) => {
+                cb && cb(null);
+                return makeFakeProjectDb({
+                    imagesRows: [imageRow],
+                    displayRow: { IName: 'image1.jpg', display_id: 1 },
+                });
+            }),
+        };
+        global.probe = jest.fn();
+        global.probe.sync = jest.fn().mockReturnValue({ width: 800, height: 600 });
+
+        await getAnnotatePage(req, res);
+
+        expect(s3Client.getObjectStream).not.toHaveBeenCalled();
+        expect(queries.managed.getBucket).not.toHaveBeenCalled();
+        expect(res.render).toHaveBeenCalledWith('annotate', expect.objectContaining({
+            image_path: 'projects/testuser-test-project/images/image1.jpg',
+            image_width: 800,
+            image_height: 600,
+        }));
+    });
+
+    it('renders 404 when there is no local file and the image is not S3-backed', async () => {
+        const imageRow = { IName: 'image1.jpg', reviewImage: 0, Source: null, SourceKey: null };
+
+        global.fs = { existsSync: jest.fn().mockReturnValue(false), readFileSync: jest.fn() };
+        global.sqlite3 = {
+            Database: jest.fn((dbPath, cb) => {
+                cb && cb(null);
+                return makeFakeProjectDb({
+                    imagesRows: [imageRow],
+                    displayRow: { IName: 'image1.jpg', display_id: 1 },
+                });
+            }),
+        };
+
+        await getAnnotatePage(req, res);
+
+        expect(queries.managed.getBucket).not.toHaveBeenCalled();
+        expect(res.render).toHaveBeenCalledWith('404', expect.any(Object));
+    });
+});
diff --git a/utils/s3Client.js b/utils/s3Client.js
index 93e90951..3668fe5e 100644
--- a/utils/s3Client.js
+++ b/utils/s3Client.js
@@ -131,9 +131,25 @@ async function downloadObjectToFile(s3Client, bucketName, key, destPath) {
     });
 }
 
+// For "stream" sync mode: fetch an object's bytes live for a single request
+// instead of persisting them to disk. Caller is responsible for piping
+// response.Body to the outgoing HTTP response.
+async function getObjectStream(s3Client, bucketName, key) {
+    const response = await s3Client.send(
+        new GetObjectCommand({ Bucket: bucketName, Key: key }),
+    );
+
+    return {
+        body: response.Body,
+        contentType: response.ContentType,
+        contentLength: response.ContentLength,
+    };
+}
+
 module.exports = {
     buildS3Client,
     verifyBucketAccess,
     listImageObjects,
     downloadObjectToFile,
+    getObjectStream,
 };
diff --git a/views/create.ejs b/views/create.ejs
index 7562b8bc..e5d890b5 100644
--- a/views/create.ejs
+++ b/views/create.ejs
@@ -184,7 +184,7 @@
             </form>
 
             <form class="add-project-form flow-content s3-flow" id="s3Form" onsubmit="return false;">
-                <h3>Add Project From S3 Bucket <span class="help-tooltip-icon help-tooltip-icon-light" data-toggle="tooltip" title="Create a project backed by an S3 bucket instead of an uploaded archive. Njobvu downloads a local working copy of the bucket's images, so existing labeling, training, and inference tools work unchanged">?</span></h3>
+                <h3>Add Project From S3 Bucket <span class="help-tooltip-icon help-tooltip-icon-light" data-toggle="tooltip" title="Create a project backed by an S3 bucket instead of an uploaded archive. Choose below whether Njobvu downloads a local working copy of the bucket's images (so existing labeling, training, and inference tools work unchanged) or streams them on demand instead">?</span></h3>
                 <div class="form-row">
                     <div class="form-group col-md-6">
                         <label for="s3_project_name">Project Name</label>
@@ -224,6 +224,15 @@
                     </div>
                 </div>
                 <p>Leave the Access Key fields blank to use the server's default AWS credentials/role.</p>
+                <div class="form-row">
+                    <div class="form-group col-md-12">
+                        <label for="s3_sync_mode">Image loading <span class="help-tooltip-icon help-tooltip-icon-light" data-toggle="tooltip" title="Download all: copies every image to local disk now, so labeling/training/inference work exactly as they do today. Stream on demand: skips the copy - images are fetched from the bucket only when actually viewed, which avoids downloading the whole bucket up front but adds a live fetch on first view of each image.">?</span></label>
+                        <select id="s3_sync_mode" name="sync_mode" aria-label="S3 image loading mode" class="form-control">
+                            <option value="download" selected>Download all images now</option>
+                            <option value="stream">Stream images on demand</option>
+                        </select>
+                    </div>
+                </div>
                 <div class="col text-center">
                     <hr class="my-4">
                     <button type="button" id="s3SubmitBtn" aria-label="submit" class="btn btn-primary btn-lg" style="border-radius: 0" onclick="submitS3Form()">submit</button>
@@ -289,6 +298,7 @@
         const accessKeyId = document.getElementById('s3_access_key_id').value;
         const secretAccessKey = document.getElementById('s3_secret_access_key').value;
         const endpoint = document.getElementById("s3_endpoint").value;
+        const syncMode = document.getElementById("s3_sync_mode").value;
 
         if (!bucketName || !region) {
             alert("Bucket Name and Region are required.");
@@ -327,6 +337,7 @@
                     AccessKeyId: accessKeyId,
                     SecretAccessKey: secretAccessKey,
                     Endpoint: endpoint,
+                    SyncMode: syncMode,
                 }),
             });
 

From bb1945b864ae9088bd6548911480e5d0bc9960a9 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:10:13 -0700
Subject: [PATCH 06/18] feat(inference): support zip extraction and S3 bucket
 image streaming up to max limit

Co-authored-by: multica-agent <github@multica.ai>
---
 controllers/inference/datatovalues.py         |   9 +-
 controllers/inference/inception.py            |  24 ++-
 controllers/inference/megadetector.py         |  32 +++-
 db/migrations.sql                             |   3 +-
 queries/s3/s3.js                              |  62 ++++--
 routes/api/v2/s3Buckets.js                    |  10 +
 routes/inference/inceptionInference.js        |  20 ++
 routes/inference/megadetectorInference.js     |  13 ++
 routes/inference/uploadInferenceFile.js       |  44 ++++-
 routes/inference/yoloInference.js             |  13 ++
 .../inferenceDatasetPipeline.test.js          | 179 ++++++++++++++++++
 utils/inferenceDatasetPipeline.js             | 102 ++++++++++
 12 files changed, 474 insertions(+), 37 deletions(-)
 create mode 100644 tests/integration/inferenceDatasetPipeline.test.js
 create mode 100644 utils/inferenceDatasetPipeline.js

diff --git a/controllers/inference/datatovalues.py b/controllers/inference/datatovalues.py
index bb1e4e68..51ae5d6a 100755
--- a/controllers/inference/datatovalues.py
+++ b/controllers/inference/datatovalues.py
@@ -275,9 +275,12 @@ def copy_raw_image(original_src: str, dest_path: str):
         err = "You need more options to run the tool"
         print(err)
 
-    if image_path == "":
-        err = "You need more options to run the tool"
-        print(err)
+    if image_path.lower().endswith(".zip") or image_path.lower().endswith(".7z") or (os.path.isfile(image_path) and zipfile.is_zipfile(image_path)):
+        import tempfile
+        temp_dir = tempfile.mkdtemp(prefix="yolo_zip_")
+        with zipfile.ZipFile(image_path, 'r') as zip_ref:
+            zip_ref.extractall(temp_dir)
+        image_path = temp_dir
 
     print("Ultralytics Version of YOLO Requested:")
 
diff --git a/controllers/inference/inception.py b/controllers/inference/inception.py
index 13fab923..2e4af891 100644
--- a/controllers/inference/inception.py
+++ b/controllers/inference/inception.py
@@ -112,10 +112,26 @@ def preprocess_image(img_path):
 # Inference
 # ---------------------------------------------------
 
-image_files = [
-    f for f in os.listdir(args.image_path)
-    if f.lower().endswith((".jpg", ".jpeg", ".png", ".bmp"))
-]
+target_dir = args.image_path
+temp_dir = None
+
+if target_dir.lower().endswith(".zip") or target_dir.lower().endswith(".7z") or zipfile.is_zipfile(target_dir):
+    import tempfile
+    import shutil
+    temp_dir = tempfile.mkdtemp(prefix="inception_zip_")
+    with zipfile.ZipFile(target_dir, 'r') as zip_ref:
+        zip_ref.extractall(temp_dir)
+    target_dir = temp_dir
+
+image_files = []
+if os.path.isdir(target_dir):
+    for root, dirs, files in os.walk(target_dir):
+        for f in files:
+            if f.lower().endswith((".jpg", ".jpeg", ".png", ".bmp", ".tif", ".tiff", ".gif", ".webp")):
+                image_files.append(os.path.relpath(os.path.join(root, f), target_dir))
+elif os.path.isfile(target_dir) and target_dir.lower().endswith((".jpg", ".jpeg", ".png", ".bmp", ".tif", ".tiff", ".gif", ".webp")):
+    image_files.append(os.path.basename(target_dir))
+    target_dir = os.path.dirname(target_dir)
 
 os.makedirs(args.output_path, exist_ok=True)
 
diff --git a/controllers/inference/megadetector.py b/controllers/inference/megadetector.py
index 7e718903..bf14afb6 100644
--- a/controllers/inference/megadetector.py
+++ b/controllers/inference/megadetector.py
@@ -266,8 +266,36 @@ def write_csvs(result, image_dir, output_dir):
 temp_dir = None
 image_dir = input_path
 
-if os.path.isdir(input_path):
+def process_directory_for_videos_and_subdirs(target_dir, fps):
+    for root, dirs, files in os.walk(target_dir):
+        for f in files:
+            full_f = os.path.join(root, f)
+            if is_video(f):
+                vid_name = os.path.splitext(f)[0]
+                frame_out = os.path.join(target_dir, f"{vid_name}_frames")
+                extract_frames(full_f, frame_out, fps)
+                for frame_f in os.listdir(frame_out):
+                    if is_image(frame_f):
+                        shutil.copy2(os.path.join(frame_out, frame_f), os.path.join(target_dir, f"{vid_name}_{frame_f}"))
+                shutil.rmtree(frame_out, ignore_errors=True)
+            elif is_image(f) and root != target_dir:
+                shutil.copy2(full_f, os.path.join(target_dir, f))
+
+if input_path.lower().endswith(".zip") or input_path.lower().endswith(".7z") or zipfile.is_zipfile(input_path):
+    if not os.path.exists(input_path):
+        print(f"Error: input archive not found: {input_path}")
+        sys.exit(1)
+    temp_dir = tempfile.mkdtemp(prefix="megadetector_zip_")
+    image_dir = os.path.join(temp_dir, "extracted")
+    os.makedirs(image_dir, exist_ok=True)
+    with zipfile.ZipFile(input_path, 'r') as zip_ref:
+        zip_ref.extractall(image_dir)
+    process_directory_for_videos_and_subdirs(image_dir, args.fps)
+
+elif os.path.isdir(input_path):
     image_dir = input_path
+    process_directory_for_videos_and_subdirs(image_dir, args.fps)
+
 elif is_video(input_path):
     if not os.path.exists(input_path):
         print(f"Error: input video not found: {input_path}")
@@ -277,6 +305,7 @@ def write_csvs(result, image_dir, output_dir):
     print(f"Extracting frames from video at {args.fps} fps...")
     extract_frames(input_path, image_dir, args.fps)
     print(f"Frames extracted to {image_dir}")
+
 elif is_image(input_path):
     if not os.path.exists(input_path):
         print(f"Error: input image not found: {input_path}")
@@ -285,6 +314,7 @@ def write_csvs(result, image_dir, output_dir):
     image_dir = os.path.join(temp_dir, "single")
     os.makedirs(image_dir, exist_ok=True)
     shutil.copy2(input_path, os.path.join(image_dir, os.path.basename(input_path)))
+
 else:
     print(f"Error: unsupported input type: {input_path}")
     sys.exit(1)
diff --git a/db/migrations.sql b/db/migrations.sql
index 864395ac..2b77f987 100644
--- a/db/migrations.sql
+++ b/db/migrations.sql
@@ -2,4 +2,5 @@ CREATE TABLE Users (Username TEXT NOT NULL PRIMARY KEY, Password TEXT NOT NULL,
 INSERT INTO Users (Username, Password, FirstName, LastName, Email) VALUES ('ZeroUser', 'Temp123', 'ZeroUser', 'ZeroUser', 'ZeroUser');
 CREATE TABLE Projects (PName VARCHAR NOT NULL, PDescription VARCHAR NOT NULL, AutoSave INTEGER NOT NULL DEFAULT 0, Admin TEXT NOT NULL DEFAULT 'ZeroUser', Validate VARCHAR NOT NULL DEFAULT 0, FOREIGN KEY(Admin) REFERENCES Users(Username), PRIMARY KEY(PName, Admin));
 CREATE TABLE Access (Username TEXT NOT NULL, PName VARCHAR NOT NULL,  Admin TEXT NOT NULL, FOREIGN KEY(Username) REFERENCES Users(Username), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
-CREATE TABLE S3Buckets (PName VARCHAR NOT NULL, Admin TEXT NOT NULL, BucketName VARCHAR NOT NULL, Region VARCHAR NOT NULL, Prefix VARCHAR NOT NULL DEFAULT '', AccessKeyId VARCHAR, SecretAccessKey VARCHAR, LastSyncedAt TEXT, Endpoint TEXT, PRIMARY KEY(PName, Admin), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
+CREATE TABLE S3Buckets (PName VARCHAR NOT NULL, Admin TEXT NOT NULL, BucketName VARCHAR NOT NULL, Region VARCHAR NOT NULL, Prefix VARCHAR NOT NULL DEFAULT '', AccessKeyId VARCHAR, SecretAccessKey VARCHAR, LastSyncedAt TEXT, Endpoint TEXT, MaxImages INTEGER DEFAULT NULL, PRIMARY KEY(PName, Admin), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
+ALTER TABLE S3Buckets ADD COLUMN MaxImages INTEGER DEFAULT NULL;
diff --git a/queries/s3/s3.js b/queries/s3/s3.js
index 5c5fe857..c3eb65e4 100644
--- a/queries/s3/s3.js
+++ b/queries/s3/s3.js
@@ -8,27 +8,53 @@ module.exports = {
             prefix,
             accessKeyId,
             secretAccessKey,
-            endpoint
+            endpoint,
+            maxImages
         ) {
-            const query =
-                "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint) " +
-                "VALUES (?, ?, ?, ?, ?, ?, ?, ?) " +
-                "ON CONFLICT(PName, Admin) DO UPDATE SET " +
-                "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
-                "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint";
+            const parsedMaxImages =
+                maxImages !== undefined && maxImages !== null && maxImages !== ""
+                    ? parseInt(maxImages, 10)
+                    : null;
 
-            const result = await global.managedDbClient.run(query, [
-                projectName,
-                admin,
-                bucketName,
-                region,
-                prefix || "",
-                accessKeyId || null,
-                secretAccessKey || null,
-                endpoint || "",
-            ]);
+            try {
+                const queryWithMax =
+                    "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint, MaxImages) " +
+                    "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?) " +
+                    "ON CONFLICT(PName, Admin) DO UPDATE SET " +
+                    "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
+                    "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint, " +
+                    "MaxImages = excluded.MaxImages";
 
-            return result;
+                return await global.managedDbClient.run(queryWithMax, [
+                    projectName,
+                    admin,
+                    bucketName,
+                    region,
+                    prefix || "",
+                    accessKeyId || null,
+                    secretAccessKey || null,
+                    endpoint || "",
+                    parsedMaxImages,
+                ]);
+            } catch (err) {
+                const queryBase =
+                    "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint) " +
+                    "VALUES (?, ?, ?, ?, ?, ?, ?, ?) " +
+                    "ON CONFLICT(PName, Admin) DO UPDATE SET " +
+                    "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
+                    "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint";
+
+                return await global.managedDbClient.run(queryBase, [
+                    projectName,
+                    admin,
+                    bucketName,
+                    region,
+                    prefix || "",
+                    accessKeyId || null,
+                    secretAccessKey || null,
+                    endpoint || "",
+                ]);
+            }
         },
         getBucket: async function(projectName, admin) {
             const query =
diff --git a/routes/api/v2/s3Buckets.js b/routes/api/v2/s3Buckets.js
index 525fa438..d09d1e23 100644
--- a/routes/api/v2/s3Buckets.js
+++ b/routes/api/v2/s3Buckets.js
@@ -38,6 +38,7 @@ async function attachS3Bucket(req, res) {
     const AccessKeyId = trimOrUndefined(body.AccessKeyId);
     const SecretAccessKey = trimOrUndefined(body.SecretAccessKey);
     const Endpoint = trimOrUndefined(body.Endpoint);
+    const MaxImages = body.MaxImages !== undefined ? body.MaxImages : body.maxImages;
 
     if (!isOwner(req, admin)) {
         return res.status(403).json({ success: false, error: "Not authorized for this project" });
@@ -71,6 +72,7 @@ async function attachS3Bucket(req, res) {
             AccessKeyId,
             SecretAccessKey,
             ...(Endpoint ? [Endpoint] : []),
+            ...(MaxImages !== undefined && MaxImages !== null ? [MaxImages] : []),
         );
 
         return res.status(200).json({ success: true });
@@ -110,6 +112,7 @@ async function getS3Bucket(req, res) {
                 Prefix: row.Prefix,
                 Endpoint: row.Endpoint,
                 LastSyncedAt: row.LastSyncedAt,
+                MaxImages: row.MaxImages || null,
                 hasCredentials: !!row.AccessKeyId,
             },
         });
@@ -159,6 +162,9 @@ async function syncS3Bucket(req, res) {
 
         console.log(bucket);
 
+        const rawMaxLimit = req.body.maxImages || req.body.MaxImages || req.body.limit || (req.query && req.query.limit) || bucket.MaxImages;
+        const maxImagesLimit = rawMaxLimit !== undefined && rawMaxLimit !== null ? parseInt(rawMaxLimit, 10) : Infinity;
+
         const s3Client = buildS3Client({
             region: bucket.Region,
             accessKeyId: bucket.AccessKeyId,
@@ -173,6 +179,10 @@ async function syncS3Bucket(req, res) {
         let skippedCount = 0;
 
         for (const key of objectKeys) {
+            if (syncedImages.length >= maxImagesLimit) {
+                break;
+            }
+
             const fileName = sanitizeFileName(path.basename(key));
 
             if (!fileName || existingImages.has(fileName)) {
diff --git a/routes/inference/inceptionInference.js b/routes/inference/inceptionInference.js
index 2d2dd10d..8d85bae9 100644
--- a/routes/inference/inceptionInference.js
+++ b/routes/inference/inceptionInference.js
@@ -1,6 +1,7 @@
 const queries = require("../../queries/queries");
 const { exec } = require("child_process");
 const formatRunOptionsHeader = require("../../utils/formatRunOptionsHeader");
+const { prepareInferenceDataset } = require("../../utils/inferenceDatasetPipeline");
 
 async function inceptionInference(req, res) {
     try {
@@ -50,6 +51,25 @@ async function inceptionInference(req, res) {
             fs.copyFileSync(inferenceScript, inferenceScriptCopyPath);
         }
 
+        const datasetResult = await prepareInferenceDataset({
+            PName,
+            Admin,
+            inference_file: inferenceFile,
+            use_s3_bucket: req.body.use_s3_bucket || req.body.s3_bucket || req.body.inference_source === "s3",
+            max_images: req.body.max_images || req.body.maxImages || req.body.limit,
+            projectPath,
+            inferenceUploadPath,
+        });
+
+        inferenceFilePath = datasetResult.inferenceFilePath || inferenceFilePath;
+
+        if (!fs.existsSync(inferenceFilePath)) {
+            const fallbackInferenceFilePath = path.join(inferenceUploadPath, inferenceFilePath);
+            if (fs.existsSync(fallbackInferenceFilePath)) {
+                inferenceFilePath = fallbackInferenceFilePath;
+            }
+        }
+
         let existingClasses;
 
         try {
diff --git a/routes/inference/megadetectorInference.js b/routes/inference/megadetectorInference.js
index 611ad991..64eba5fc 100644
--- a/routes/inference/megadetectorInference.js
+++ b/routes/inference/megadetectorInference.js
@@ -1,6 +1,7 @@
 const path = require("path");
 const formatRunOptionsHeader = require("../../utils/formatRunOptionsHeader");
 const config = require("../../utils/config");
+const { prepareInferenceDataset } = require("../../utils/inferenceDatasetPipeline");
 
 async function megadetectorInference(req, res) {
     try {
@@ -45,6 +46,18 @@ async function megadetectorInference(req, res) {
             fs.copyFileSync(megadetectorScript, megadetectorScriptCopyPath);
         }
 
+        const datasetResult = await prepareInferenceDataset({
+            PName,
+            Admin,
+            inference_file: inferenceFile,
+            use_s3_bucket: req.body.use_s3_bucket || req.body.s3_bucket || req.body.inference_source === "s3",
+            max_images: req.body.max_images || req.body.maxImages || req.body.limit,
+            projectPath,
+            inferenceUploadPath,
+        });
+
+        inferenceFilePath = datasetResult.inferenceFilePath || inferenceFilePath;
+
         if (!fs.existsSync(inferenceFilePath)) {
             const fallbackInferenceFilePath = path.join(inferenceUploadPath, inferenceFilePath);
             if (fs.existsSync(fallbackInferenceFilePath)) {
diff --git a/routes/inference/uploadInferenceFile.js b/routes/inference/uploadInferenceFile.js
index 0e894bcd..5ab4bb7a 100644
--- a/routes/inference/uploadInferenceFile.js
+++ b/routes/inference/uploadInferenceFile.js
@@ -1,37 +1,61 @@
+const path = require("path");
+const fs = require("fs");
+const unzipFile = require("../../utils/unzipFile");
+
 async function uploadInferenceFile(req, res) {
     var PName = req.body.PName,
         Admin = req.body.Admin,
-        user = req.cookies.Username,
-        inferenceFile = req.files.upload_inference;
+        user = req.cookies && req.cookies.Username,
+        inferenceFile = req.files && req.files.upload_inference;
+
+    if (!inferenceFile) {
+        return res.status(400).send({ Success: "ERROR: No file uploaded" });
+    }
 
     var publicPath = currentPath,
-        mainPath = publicPath + "public/projects/", // $LABELING_TOOL_PATH/public/projects/
-        projectPath = mainPath + Admin + "-" + PName, // $LABELING_TOOL_PATH/public/projects/project_name
-        trainingPath = projectPath + "/training",
+        mainPath = publicPath + "public/projects/",
+        projectPath = mainPath + Admin + "-" + PName,
         inferencePath = projectPath + "/inference/",
         inferenceUploadPath = inferencePath + "/uploads/",
         inferenceFilePath = inferenceUploadPath + inferenceFile.name;
 
-    const validFileNames = ["png", "tif", "jpg", "jpeg", "gif", "mp4", "mov"];
+    const validFileNames = ["png", "tif", "jpg", "jpeg", "gif", "mp4", "mov", "zip", "7z"];
+    const ext = inferenceFile.name.split(".").pop().toLowerCase();
 
-    if (!validFileNames.includes(inferenceFile.name.split(".").pop())) {
+    if (!validFileNames.includes(ext)) {
         res.send({
             Success:
-                "ERROR: Wrong filetype. Must be type .png, .jpg, jpeg, tif or .gif",
+                "ERROR: Wrong filetype. Must be type .png, .jpg, jpeg, tif, .gif, .mp4, .mov, .zip, or .7z",
         });
     } else {
         if (!fs.existsSync(inferencePath)) {
-            fs.mkdirSync(inferencePath);
+            fs.mkdirSync(inferencePath, { recursive: true });
         }
 
         if (!fs.existsSync(inferenceUploadPath)) {
-            fs.mkdir(inferenceUploadPath);
+            fs.mkdirSync(inferenceUploadPath, { recursive: true });
         }
 
         await inferenceFile.mv(inferenceFilePath);
 
+        let extractedPath = null;
+        if (ext === "zip" || ext === "7z") {
+            const folderName = path.parse(inferenceFile.name).name;
+            const outputDir = path.join(inferenceUploadPath, folderName);
+            try {
+                await unzipFile(inferenceFilePath, outputDir);
+                extractedPath = outputDir;
+            } catch (err) {
+                if (global.logger) {
+                    global.logger.error(err);
+                }
+            }
+        }
+
         res.send({
             Success: "Your inference file has been uploaded and saved",
+            filename: inferenceFile.name,
+            extractedPath,
         });
     }
 }
diff --git a/routes/inference/yoloInference.js b/routes/inference/yoloInference.js
index 3e6c9ffc..813b5acb 100644
--- a/routes/inference/yoloInference.js
+++ b/routes/inference/yoloInference.js
@@ -2,6 +2,7 @@ const queries = require("../../queries/queries");
 const path = require("path");
 const formatRunOptionsHeader = require("../../utils/formatRunOptionsHeader");
 const config = require("../../utils/config");
+const { prepareInferenceDataset } = require("../../utils/inferenceDatasetPipeline");
 
 async function yoloInference(req, res) {
     try {
@@ -149,6 +150,18 @@ async function yoloInference(req, res) {
 
         ultralyticsProjectRun = runPath;
 
+        const datasetResult = await prepareInferenceDataset({
+            PName,
+            Admin,
+            inference_file: inferenceFile,
+            use_s3_bucket: req.body.use_s3_bucket || req.body.s3_bucket || req.body.inference_source === "s3",
+            max_images: req.body.max_images || req.body.maxImages || req.body.limit,
+            projectPath,
+            inferenceUploadPath,
+        });
+
+        inferenceFilePath = datasetResult.inferenceFilePath || inferenceFilePath;
+
         if (!fs.existsSync(inferenceFilePath)) {
             const fallbackInferenceFilePath = path.join(inferenceUploadPath, inferenceFilePath);
             if (fs.existsSync(fallbackInferenceFilePath)) {
diff --git a/tests/integration/inferenceDatasetPipeline.test.js b/tests/integration/inferenceDatasetPipeline.test.js
new file mode 100644
index 00000000..2f724280
--- /dev/null
+++ b/tests/integration/inferenceDatasetPipeline.test.js
@@ -0,0 +1,179 @@
+const fs = require("fs");
+const path = require("path");
+const request = require("supertest");
+const express = require("express");
+const cookieParser = require("cookie-parser");
+
+const { prepareInferenceDataset } = require("../../utils/inferenceDatasetPipeline");
+const uploadInferenceFile = require("../../routes/inference/uploadInferenceFile");
+const { attachS3Bucket, syncS3Bucket } = require("../../routes/api/v2/s3Buckets");
+const queries = require("../../queries/queries");
+const s3Client = require("../../utils/s3Client");
+
+jest.mock("../../queries/queries", () => ({
+    managed: {
+        getBucket: jest.fn(),
+        attachBucket: jest.fn().mockResolvedValue({ success: true }),
+        touchBucketSyncedAt: jest.fn().mockResolvedValue({ success: true }),
+    },
+    project: {
+        addImages: jest.fn().mockResolvedValue({ success: true }),
+    },
+}));
+
+jest.mock("../../utils/s3Client", () => ({
+    buildS3Client: jest.fn(() => ({ fake: true })),
+    verifyBucketAccess: jest.fn().mockResolvedValue(undefined),
+    listImageObjects: jest.fn().mockResolvedValue(["img1.jpg", "img2.jpg", "img3.jpg"]),
+    downloadObjectToFile: jest.fn().mockImplementation(async (client, bucket, key, dest) => {
+        require("fs").writeFileSync(dest, "fake image content");
+    }),
+}));
+
+jest.mock("../../utils/unzipFile", () => jest.fn().mockImplementation(async (zipPath, outDir) => {
+    const mockFs = require("fs");
+    mockFs.mkdirSync(outDir, { recursive: true });
+    mockFs.writeFileSync(require("path").join(outDir, "extracted_image.jpg"), "extracted content");
+    mockFs.writeFileSync(require("path").join(outDir, "extracted_video.mp4"), "video content");
+}));
+
+function buildApp() {
+    const app = express();
+    app.use(express.json());
+    app.use(cookieParser());
+    app.post("/upload_inference_file", (req, res, next) => {
+        req.files = req.files || {};
+        next();
+    }, uploadInferenceFile);
+    app.post("/api/v2/projects/:admin/:projectName/s3-bucket", attachS3Bucket);
+    app.post("/api/v2/projects/:admin/:projectName/s3-bucket/sync", syncS3Bucket);
+    return app;
+}
+
+describe("Inference Dataset Pipeline & S3 Max Image Limit", () => {
+    const tmpDir = path.join(__dirname, "../tmp_dataset_test");
+
+    beforeAll(() => {
+        global.currentPath = tmpDir + "/";
+        global.logger = { error: jest.fn(), debug: jest.fn() };
+        global.readdirAsync = jest.fn().mockResolvedValue([]);
+        fs.mkdirSync(tmpDir, { recursive: true });
+    });
+
+    afterAll(() => {
+        if (fs.existsSync(tmpDir)) {
+            fs.rmSync(tmpDir, { recursive: true, force: true });
+        }
+    });
+
+    afterEach(() => {
+        jest.clearAllMocks();
+    });
+
+    describe("prepareInferenceDataset utility", () => {
+        it("extracts zip archives (images & videos) for inference", async () => {
+            const uploadDir = path.join(tmpDir, "public/projects/admin1-proj1/inference/uploads");
+            fs.mkdirSync(uploadDir, { recursive: true });
+            const zipPath = path.join(uploadDir, "sample.zip");
+            fs.writeFileSync(zipPath, "dummy zip binary");
+
+            const result = await prepareInferenceDataset({
+                PName: "proj1",
+                Admin: "admin1",
+                inference_file: "sample.zip",
+                projectPath: path.join(tmpDir, "public/projects/admin1-proj1"),
+                inferenceUploadPath: uploadDir,
+            });
+
+            expect(result.isZipExtracted).toBe(true);
+            expect(result.inferenceFilePath).toBe(path.join(uploadDir, "sample"));
+            expect(fs.existsSync(path.join(uploadDir, "sample/extracted_image.jpg"))).toBe(true);
+        });
+
+        it("pulls S3 image stream up to configured maxImages limit for inference", async () => {
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: {
+                    BucketName: "my-bucket",
+                    Region: "us-east-1",
+                    Prefix: "images/",
+                    MaxImages: 2,
+                    AccessKeyId: "key",
+                    SecretAccessKey: "secret",
+                },
+            });
+
+            const uploadDir = path.join(tmpDir, "public/projects/admin1-proj1/inference/uploads");
+            fs.mkdirSync(uploadDir, { recursive: true });
+
+            const result = await prepareInferenceDataset({
+                PName: "proj1",
+                Admin: "admin1",
+                use_s3_bucket: true,
+                max_images: 2,
+                projectPath: path.join(tmpDir, "public/projects/admin1-proj1"),
+                inferenceUploadPath: uploadDir,
+            });
+
+            expect(result.isS3Stream).toBe(true);
+            expect(result.syncedCount).toBe(2);
+            expect(s3Client.downloadObjectToFile).toHaveBeenCalledTimes(2);
+        });
+    });
+
+    describe("syncS3Bucket with maxImages limit", () => {
+        it("respects maxImages limit during bucket sync", async () => {
+            const projectDir = path.join(tmpDir, "public/projects/testuser-test-project/images");
+            fs.mkdirSync(projectDir, { recursive: true });
+
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: {
+                    BucketName: "test-bucket",
+                    Region: "us-east-1",
+                    Prefix: "images/",
+                    AccessKeyId: "key",
+                    SecretAccessKey: "secret",
+                },
+            });
+            s3Client.listImageObjects.mockResolvedValueOnce(["images/1.jpg", "images/2.jpg", "images/3.jpg"]);
+
+            const app = buildApp();
+            const res = await request(app)
+                .post("/api/v2/projects/testuser/test-project/s3-bucket/sync")
+                .set("Cookie", ["Username=testuser"])
+                .send({ maxImages: 2 });
+
+            expect(res.statusCode).toBe(200);
+            expect(res.body.syncedCount).toBe(2);
+            expect(s3Client.downloadObjectToFile).toHaveBeenCalledTimes(2);
+        });
+    });
+
+    describe("attachS3Bucket with MaxImages field", () => {
+        it("saves MaxImages when configuring bucket", async () => {
+            const projectDir = path.join(tmpDir, "public/projects/testuser-test-project");
+            fs.mkdirSync(projectDir, { recursive: true });
+
+            const app = buildApp();
+            const res = await request(app)
+                .post("/api/v2/projects/testuser/test-project/s3-bucket")
+                .set("Cookie", ["Username=testuser"])
+                .send({
+                    BucketName: "test-bucket",
+                    Region: "us-east-1",
+                    MaxImages: 50,
+                });
+
+            expect(res.statusCode).toBe(200);
+            expect(queries.managed.attachBucket).toHaveBeenCalledWith(
+                "test-project",
+                "testuser",
+                "test-bucket",
+                "us-east-1",
+                "",
+                undefined,
+                undefined,
+                50,
+            );
+        });
+    });
+});
diff --git a/utils/inferenceDatasetPipeline.js b/utils/inferenceDatasetPipeline.js
new file mode 100644
index 00000000..daef23fe
--- /dev/null
+++ b/utils/inferenceDatasetPipeline.js
@@ -0,0 +1,102 @@
+const fs = require("fs");
+const path = require("path");
+const unzipFile = require("./unzipFile");
+const queries = require("../queries/queries");
+const { buildS3Client, listImageObjects, downloadObjectToFile } = require("./s3Client");
+
+async function prepareInferenceDataset(options) {
+    const {
+        PName,
+        Admin,
+        inference_file,
+        use_s3_bucket,
+        max_images,
+        maxImages,
+        limit,
+        projectPath,
+        inferenceUploadPath,
+    } = options;
+
+    const effectiveMaxImages = max_images || maxImages || limit;
+
+    let targetFilePath = inference_file;
+
+    // S3 Bucket stream mode
+    if (use_s3_bucket || inference_file === "s3" || inference_file === "s3_bucket") {
+        const bucketResult = await queries.managed.getBucket(PName, Admin);
+        const bucket = bucketResult && bucketResult.row;
+
+        if (!bucket) {
+            throw new Error("No S3 bucket attached to this project");
+        }
+
+        const rawLimit = effectiveMaxImages || bucket.MaxImages || 100;
+        const maxLimit = parseInt(rawLimit, 10);
+        const dateStamp = Date.now();
+        const s3StreamFolder = path.join(inferenceUploadPath, `s3_stream_${dateStamp}`);
+
+        if (!fs.existsSync(s3StreamFolder)) {
+            fs.mkdirSync(s3StreamFolder, { recursive: true });
+        }
+
+        const s3Client = buildS3Client({
+            region: bucket.Region,
+            accessKeyId: bucket.AccessKeyId,
+            secretAccessKey: bucket.SecretAccessKey,
+            endpoint: bucket.Endpoint,
+        });
+
+        const objectKeys = await listImageObjects(s3Client, bucket.BucketName, bucket.Prefix);
+        let downloaded = 0;
+
+        for (const key of objectKeys) {
+            if (downloaded >= maxLimit) {
+                break;
+            }
+            const fileName = path.basename(key).replace(/[ +]/g, "_");
+            const destPath = path.join(s3StreamFolder, fileName);
+            await downloadObjectToFile(s3Client, bucket.BucketName, key, destPath);
+            downloaded++;
+        }
+
+        return {
+            inferenceFilePath: s3StreamFolder,
+            isS3Stream: true,
+            syncedCount: downloaded,
+        };
+    }
+
+    // Resolve local file path
+    if (targetFilePath && !fs.existsSync(targetFilePath)) {
+        const fallbackPath = path.join(inferenceUploadPath, targetFilePath);
+        if (fs.existsSync(fallbackPath)) {
+            targetFilePath = fallbackPath;
+        }
+    }
+
+    // Zip/7z archive extraction
+    if (targetFilePath && fs.existsSync(targetFilePath)) {
+        const ext = path.extname(targetFilePath).toLowerCase();
+        if (ext === ".zip" || ext === ".7z") {
+            const folderName = path.parse(targetFilePath).name;
+            const outputDir = path.join(inferenceUploadPath, folderName);
+
+            if (!fs.existsSync(outputDir)) {
+                await unzipFile(targetFilePath, outputDir);
+            }
+
+            return {
+                inferenceFilePath: outputDir,
+                isZipExtracted: true,
+            };
+        }
+    }
+
+    return {
+        inferenceFilePath: targetFilePath || inference_file,
+    };
+}
+
+module.exports = {
+    prepareInferenceDataset,
+};

From 6a87d111b3c5d6db6d5d8c81318bb219141caf4e Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:10:47 -0700
Subject: [PATCH 07/18] feat: add .zip file support for inference and S3 bucket
 config form with max image limit

Co-authored-by: multica-agent <github@multica.ai>
---
 db/migrations.sql                            |   3 +-
 queries/s3/s3.js                             |  15 +-
 routes/api/v2/s3Buckets.js                   |  16 +-
 routes/inference/uploadInferenceFile.js      |  29 +++-
 tests/integration/s3Buckets.test.js          |  26 +++
 tests/integration/uploadInferenceZip.test.js |  70 ++++++++
 views/includes/s3BucketForm.ejs              | 171 +++++++++++++++++++
 views/settings/imagesSettings.ejs            |  12 +-
 views/training/inceptionSettings.ejs         |   8 +-
 views/training/megadetectorSettings.ejs      |   6 +-
 views/training/yolovXInferenceSettings.ejs   |   7 +-
 11 files changed, 339 insertions(+), 24 deletions(-)
 create mode 100644 tests/integration/uploadInferenceZip.test.js
 create mode 100644 views/includes/s3BucketForm.ejs

diff --git a/db/migrations.sql b/db/migrations.sql
index 864395ac..b18466e2 100644
--- a/db/migrations.sql
+++ b/db/migrations.sql
@@ -2,4 +2,5 @@ CREATE TABLE Users (Username TEXT NOT NULL PRIMARY KEY, Password TEXT NOT NULL,
 INSERT INTO Users (Username, Password, FirstName, LastName, Email) VALUES ('ZeroUser', 'Temp123', 'ZeroUser', 'ZeroUser', 'ZeroUser');
 CREATE TABLE Projects (PName VARCHAR NOT NULL, PDescription VARCHAR NOT NULL, AutoSave INTEGER NOT NULL DEFAULT 0, Admin TEXT NOT NULL DEFAULT 'ZeroUser', Validate VARCHAR NOT NULL DEFAULT 0, FOREIGN KEY(Admin) REFERENCES Users(Username), PRIMARY KEY(PName, Admin));
 CREATE TABLE Access (Username TEXT NOT NULL, PName VARCHAR NOT NULL,  Admin TEXT NOT NULL, FOREIGN KEY(Username) REFERENCES Users(Username), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
-CREATE TABLE S3Buckets (PName VARCHAR NOT NULL, Admin TEXT NOT NULL, BucketName VARCHAR NOT NULL, Region VARCHAR NOT NULL, Prefix VARCHAR NOT NULL DEFAULT '', AccessKeyId VARCHAR, SecretAccessKey VARCHAR, LastSyncedAt TEXT, Endpoint TEXT, PRIMARY KEY(PName, Admin), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
+CREATE TABLE S3Buckets (PName VARCHAR NOT NULL, Admin TEXT NOT NULL, BucketName VARCHAR NOT NULL, Region VARCHAR NOT NULL, Prefix VARCHAR NOT NULL DEFAULT '', AccessKeyId VARCHAR, SecretAccessKey VARCHAR, LastSyncedAt TEXT, Endpoint TEXT, MaxImages INTEGER, PRIMARY KEY(PName, Admin), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
+ALTER TABLE S3Buckets ADD COLUMN MaxImages INTEGER;
diff --git a/queries/s3/s3.js b/queries/s3/s3.js
index 5c5fe857..d9f036a0 100644
--- a/queries/s3/s3.js
+++ b/queries/s3/s3.js
@@ -8,14 +8,20 @@ module.exports = {
             prefix,
             accessKeyId,
             secretAccessKey,
-            endpoint
+            endpoint,
+            maxImages
         ) {
+            const parsedMax = (maxImages !== undefined && maxImages !== null && maxImages !== "")
+                ? parseInt(maxImages, 10)
+                : null;
+            const validMax = Number.isFinite(parsedMax) && parsedMax > 0 ? parsedMax : null;
+
             const query =
-                "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint) " +
-                "VALUES (?, ?, ?, ?, ?, ?, ?, ?) " +
+                "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint, MaxImages) " +
+                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?) " +
                 "ON CONFLICT(PName, Admin) DO UPDATE SET " +
                 "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
-                "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint";
+                "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint, MaxImages = excluded.MaxImages";
 
             const result = await global.managedDbClient.run(query, [
                 projectName,
@@ -26,6 +32,7 @@ module.exports = {
                 accessKeyId || null,
                 secretAccessKey || null,
                 endpoint || "",
+                validMax,
             ]);
 
             return result;
diff --git a/routes/api/v2/s3Buckets.js b/routes/api/v2/s3Buckets.js
index 525fa438..4a6736e4 100644
--- a/routes/api/v2/s3Buckets.js
+++ b/routes/api/v2/s3Buckets.js
@@ -38,6 +38,8 @@ async function attachS3Bucket(req, res) {
     const AccessKeyId = trimOrUndefined(body.AccessKeyId);
     const SecretAccessKey = trimOrUndefined(body.SecretAccessKey);
     const Endpoint = trimOrUndefined(body.Endpoint);
+    const rawMaxImages = body.MaxImages !== undefined ? body.MaxImages : (body.maxImages !== undefined ? body.maxImages : body.max_images);
+    const MaxImages = (rawMaxImages !== undefined && rawMaxImages !== null && rawMaxImages !== "") ? parseInt(rawMaxImages, 10) : null;
 
     if (!isOwner(req, admin)) {
         return res.status(403).json({ success: false, error: "Not authorized for this project" });
@@ -70,7 +72,8 @@ async function attachS3Bucket(req, res) {
             Prefix || "",
             AccessKeyId,
             SecretAccessKey,
-            ...(Endpoint ? [Endpoint] : []),
+            Endpoint || "",
+            MaxImages
         );
 
         return res.status(200).json({ success: true });
@@ -102,6 +105,8 @@ async function getS3Bucket(req, res) {
             return res.status(404).json({ success: false, error: "No S3 bucket attached" });
         }
 
+        const maxImg = row.MaxImages != null ? Number(row.MaxImages) : null;
+
         return res.status(200).json({
             success: true,
             bucket: {
@@ -109,6 +114,8 @@ async function getS3Bucket(req, res) {
                 Region: row.Region,
                 Prefix: row.Prefix,
                 Endpoint: row.Endpoint,
+                MaxImages: maxImg,
+                max_images: maxImg,
                 LastSyncedAt: row.LastSyncedAt,
                 hasCredentials: !!row.AccessKeyId,
             },
@@ -157,8 +164,6 @@ async function syncS3Bucket(req, res) {
             return res.status(404).json({ success: false, error: "No S3 bucket attached to this project" });
         }
 
-        console.log(bucket);
-
         const s3Client = buildS3Client({
             region: bucket.Region,
             accessKeyId: bucket.AccessKeyId,
@@ -166,6 +171,7 @@ async function syncS3Bucket(req, res) {
             endpoint: bucket.Endpoint,
         });
 
+        const maxImages = bucket.MaxImages != null ? Number(bucket.MaxImages) : null;
         const objectKeys = await listImageObjects(s3Client, bucket.BucketName, bucket.Prefix);
         const existingImages = new Set(await global.readdirAsync(imagesPath));
 
@@ -173,6 +179,10 @@ async function syncS3Bucket(req, res) {
         let skippedCount = 0;
 
         for (const key of objectKeys) {
+            if (Number.isFinite(maxImages) && maxImages > 0 && existingImages.size >= maxImages) {
+                break;
+            }
+
             const fileName = sanitizeFileName(path.basename(key));
 
             if (!fileName || existingImages.has(fileName)) {
diff --git a/routes/inference/uploadInferenceFile.js b/routes/inference/uploadInferenceFile.js
index 0e894bcd..cdb23aea 100644
--- a/routes/inference/uploadInferenceFile.js
+++ b/routes/inference/uploadInferenceFile.js
@@ -1,3 +1,5 @@
+const fs = require("fs");
+
 async function uploadInferenceFile(req, res) {
     var PName = req.body.PName,
         Admin = req.body.Admin,
@@ -12,23 +14,38 @@ async function uploadInferenceFile(req, res) {
         inferenceUploadPath = inferencePath + "/uploads/",
         inferenceFilePath = inferenceUploadPath + inferenceFile.name;
 
-    const validFileNames = ["png", "tif", "jpg", "jpeg", "gif", "mp4", "mov"];
+    const validFileNames = ["png", "tif", "jpg", "jpeg", "gif", "mp4", "mov", "zip"];
+    const fileExt = (inferenceFile.name.split(".").pop() || "").toLowerCase();
 
-    if (!validFileNames.includes(inferenceFile.name.split(".").pop())) {
+    if (!validFileNames.includes(fileExt)) {
         res.send({
             Success:
-                "ERROR: Wrong filetype. Must be type .png, .jpg, jpeg, tif or .gif",
+                "ERROR: Wrong filetype. Must be type .png, .jpg, .jpeg, .tif, .gif, .mp4, .mov, or .zip",
         });
     } else {
         if (!fs.existsSync(inferencePath)) {
-            fs.mkdirSync(inferencePath);
+            fs.mkdirSync(inferencePath, { recursive: true });
         }
 
         if (!fs.existsSync(inferenceUploadPath)) {
-            fs.mkdir(inferenceUploadPath);
+            fs.mkdirSync(inferenceUploadPath, { recursive: true });
         }
 
-        await inferenceFile.mv(inferenceFilePath);
+        if (inferenceFile && typeof inferenceFile.mv === "function") {
+            await new Promise((resolve, reject) => {
+                try {
+                    const ret = inferenceFile.mv(inferenceFilePath, (err) => {
+                        if (err) return reject(err);
+                        resolve();
+                    });
+                    if (ret && typeof ret.then === "function") {
+                        ret.then(resolve).catch(reject);
+                    }
+                } catch (e) {
+                    reject(e);
+                }
+            });
+        }
 
         res.send({
             Success: "Your inference file has been uploaded and saved",
diff --git a/tests/integration/s3Buckets.test.js b/tests/integration/s3Buckets.test.js
index 94d7caab..d8c0bd0d 100644
--- a/tests/integration/s3Buckets.test.js
+++ b/tests/integration/s3Buckets.test.js
@@ -96,6 +96,32 @@ describe('S3 Bucket Routes', () => {
                 'images/',
                 'AKIA...',
                 'secret',
+                '',
+                null,
+            );
+        });
+
+        it('attaches a bucket with MaxImages limit', async () => {
+            const res = await request(app)
+                .post('/api/v2/projects/testuser/test-project/s3-bucket')
+                .set('Cookie', ['Username=testuser'])
+                .send({
+                    ...validBody,
+                    MaxImages: 50,
+                });
+
+            expect(res.statusCode).toBe(200);
+            expect(res.body.success).toBe(true);
+            expect(queries.managed.attachBucket).toHaveBeenCalledWith(
+                'test-project',
+                'testuser',
+                'my-bucket',
+                'us-east-1',
+                'images/',
+                'AKIA...',
+                'secret',
+                '',
+                50,
             );
         });
 
diff --git a/tests/integration/uploadInferenceZip.test.js b/tests/integration/uploadInferenceZip.test.js
new file mode 100644
index 00000000..50c427bd
--- /dev/null
+++ b/tests/integration/uploadInferenceZip.test.js
@@ -0,0 +1,70 @@
+const fs = require('fs');
+const path = require('path');
+const uploadInferenceFile = require('../../routes/inference/uploadInferenceFile');
+
+describe('uploadInferenceFile Controller', () => {
+    const testDir = path.join(__dirname, '../tmp_upload_test');
+
+    beforeAll(() => {
+        global.currentPath = testDir + '/';
+        if (!fs.existsSync(testDir)) {
+            fs.mkdirSync(testDir, { recursive: true });
+        }
+    });
+
+    afterAll(() => {
+        if (fs.existsSync(testDir)) {
+            fs.rmSync(testDir, { recursive: true, force: true });
+        }
+    });
+
+    function mockReqRes(fileName) {
+        const req = {
+            body: { PName: 'test-project', Admin: 'testuser' },
+            cookies: { Username: 'testuser' },
+            files: {
+                upload_inference: {
+                    name: fileName,
+                    mv: jest.fn().mockImplementation((dest, cb) => {
+                        if (typeof cb === 'function') cb(null);
+                        return Promise.resolve();
+                    }),
+                },
+            },
+        };
+        const res = {
+            send: jest.fn(),
+        };
+        return { req, res };
+    }
+
+    it('accepts .zip files for inference dataset', async () => {
+        const { req, res } = mockReqRes('images_archive.zip');
+        await uploadInferenceFile(req, res);
+
+        expect(res.send).toHaveBeenCalledWith({
+            Success: 'Your inference file has been uploaded and saved',
+        });
+        expect(req.files.upload_inference.mv).toHaveBeenCalled();
+    });
+
+    it('accepts valid image/video file extensions (.jpg, .mp4, .mov)', async () => {
+        const { req, res } = mockReqRes('video_sample.mp4');
+        await uploadInferenceFile(req, res);
+
+        expect(res.send).toHaveBeenCalledWith({
+            Success: 'Your inference file has been uploaded and saved',
+        });
+        expect(req.files.upload_inference.mv).toHaveBeenCalled();
+    });
+
+    it('rejects unsupported file extensions (.exe, .txt)', async () => {
+        const { req, res } = mockReqRes('malicious.exe');
+        await uploadInferenceFile(req, res);
+
+        expect(res.send).toHaveBeenCalledWith({
+            Success: 'ERROR: Wrong filetype. Must be type .png, .jpg, .jpeg, .tif, .gif, .mp4, .mov, or .zip',
+        });
+        expect(req.files.upload_inference.mv).not.toHaveBeenCalled();
+    });
+});
diff --git a/views/includes/s3BucketForm.ejs b/views/includes/s3BucketForm.ejs
new file mode 100644
index 00000000..e703111d
--- /dev/null
+++ b/views/includes/s3BucketForm.ejs
@@ -0,0 +1,171 @@
+<div class="card mb-4 mt-3" id="s3BucketSection">
+    <div class="card-header bg-dark text-white d-flex justify-content-between align-items-center">
+        <h5 class="mb-0">S3 Bucket Configuration (Inference)</h5>
+        <button class="btn btn-sm btn-outline-light" type="button" data-toggle="collapse" data-target="#s3BucketFormCollapse" aria-expanded="false" aria-controls="s3BucketFormCollapse">
+            Attach / Configure S3 Bucket
+        </button>
+    </div>
+    <div class="collapse" id="s3BucketFormCollapse">
+        <div class="card-body">
+            <p class="text-muted">
+                Attach an S3 bucket to stream or sync images for inference. You can configure a maximum image limit to restrict the number of pulled images.
+            </p>
+            <div id="s3BucketCurrentStatus" class="alert alert-info d-none mb-3"></div>
+            <form id="s3BucketForm" onsubmit="return false;">
+                <div class="form-row">
+                    <div class="form-group col-md-6">
+                        <label for="s3_bucket_name">Bucket Name <span class="text-danger">*</span></label>
+                        <input id="s3_bucket_name" class="form-control" type="text" placeholder="my-image-bucket" required>
+                    </div>
+                    <div class="form-group col-md-6">
+                        <label for="s3_region">Region <span class="text-danger">*</span></label>
+                        <input id="s3_region" class="form-control" type="text" placeholder="us-east-1" required>
+                    </div>
+                </div>
+                <div class="form-row">
+                    <div class="form-group col-md-4">
+                        <label for="s3_prefix">Prefix (Optional)</label>
+                        <input id="s3_prefix" class="form-control" type="text" placeholder="images/">
+                    </div>
+                    <div class="form-group col-md-4">
+                        <label for="s3_endpoint">S3 Endpoint (Optional)</label>
+                        <input id="s3_endpoint" type="text" class="form-control" placeholder="https://s3.amazonaws.com">
+                    </div>
+                    <div class="form-group col-md-4">
+                        <label for="s3_max_images">Max Image Limit (Optional)</label>
+                        <input id="s3_max_images" name="max_images" type="number" min="1" class="form-control" placeholder="Unlimited (e.g. 100)">
+                    </div>
+                </div>
+                <div class="form-row">
+                    <div class="form-group col-md-6">
+                        <label for="s3_access_key_id">Access Key ID</label>
+                        <input id="s3_access_key_id" class="form-control" type="text" autocomplete="off">
+                    </div>
+                    <div class="form-group col-md-6">
+                        <label for="s3_secret_access_key">Secret Access Key</label>
+                        <input id="s3_secret_access_key" class="form-control" type="password" autocomplete="off">
+                    </div>
+                </div>
+                <div>
+                    <label class="mb-2 text-info" id="s3progress" aria-live="polite" aria-atomic="true"></label>
+                </div>
+                <button class="btn btn-primary" id="s3SubmitBtn" type="button" onclick="attachAndSyncS3Bucket()">Attach &amp; Sync Bucket</button>
+                <button class="btn btn-danger d-none ml-2" id="s3DetachBtn" type="button" onclick="detachS3Bucket()">Detach Bucket</button>
+            </form>
+        </div>
+    </div>
+</div>
+
+<script>
+    async function loadS3BucketStatus() {
+        try {
+            const res = await fetch(`/api/v2/projects/<%= Admin %>/<%= PName %>/s3-bucket`);
+            if (res.ok) {
+                const data = await res.json();
+                if (data.success && data.bucket) {
+                    const b = data.bucket;
+                    const statusEl = document.getElementById('s3BucketCurrentStatus');
+                    if (statusEl) {
+                        statusEl.classList.remove('d-none');
+                        let limitText = b.MaxImages ? ` (Max Images: ${b.MaxImages})` : ' (Unlimited images)';
+                        statusEl.innerHTML = `<strong>Attached S3 Bucket:</strong> <code>${b.BucketName}</code> [Region: <code>${b.Region}</code>]${limitText}`;
+                    }
+                    if (document.getElementById('s3_bucket_name')) document.getElementById('s3_bucket_name').value = b.BucketName || '';
+                    if (document.getElementById('s3_region')) document.getElementById('s3_region').value = b.Region || '';
+                    if (document.getElementById('s3_prefix')) document.getElementById('s3_prefix').value = b.Prefix || '';
+                    if (document.getElementById('s3_endpoint')) document.getElementById('s3_endpoint').value = b.Endpoint || '';
+                    if (b.MaxImages && document.getElementById('s3_max_images')) {
+                        document.getElementById('s3_max_images').value = b.MaxImages;
+                    }
+                    const detachBtn = document.getElementById('s3DetachBtn');
+                    if (detachBtn) detachBtn.classList.remove('d-none');
+                }
+            }
+        } catch (e) {
+            console.log('No S3 bucket attached or error loading bucket status', e);
+        }
+    }
+
+    async function attachAndSyncS3Bucket() {
+        const bucketName = document.getElementById('s3_bucket_name').value;
+        const region = document.getElementById('s3_region').value;
+        const prefix = document.getElementById('s3_prefix').value;
+        const accessKeyId = document.getElementById('s3_access_key_id').value;
+        const secretAccessKey = document.getElementById('s3_secret_access_key').value;
+        const endpoint = document.getElementById("s3_endpoint").value;
+        const maxImages = document.getElementById("s3_max_images").value;
+
+        if (!bucketName || !region) {
+            alert("Bucket Name and Region are required.");
+            return;
+        }
+
+        const submitBtn = document.getElementById('s3SubmitBtn');
+        const progress = document.getElementById('s3progress');
+        submitBtn.disabled = true;
+
+        try {
+            progress.textContent = "Attaching S3 bucket...";
+
+            const attachRes = await fetch(`/api/v2/projects/<%= Admin %>/<%= PName %>/s3-bucket`, {
+                method: 'POST',
+                headers: { 'Content-Type': 'application/json' },
+                body: JSON.stringify({
+                    BucketName: bucketName,
+                    Region: region,
+                    Prefix: prefix,
+                    AccessKeyId: accessKeyId,
+                    SecretAccessKey: secretAccessKey,
+                    Endpoint: endpoint,
+                    MaxImages: maxImages ? parseInt(maxImages, 10) : null,
+                }),
+            });
+
+            const attachBody = await attachRes.json();
+            if (!attachRes.ok || !attachBody.success) {
+                throw new Error(attachBody.error || "Could not attach S3 bucket");
+            }
+
+            progress.textContent = "Syncing images from S3...";
+
+            const syncRes = await fetch(`/api/v2/projects/<%= Admin %>/<%= PName %>/s3-bucket/sync`, {
+                method: 'POST',
+            });
+
+            const syncBody = await syncRes.json();
+            if (!syncRes.ok || !syncBody.success) {
+                throw new Error(syncBody.error || "Could not sync images from S3 bucket");
+            }
+
+            progress.textContent = "";
+            alert(`Synced ${syncBody.syncedCount} new image(s) from S3 (skipped ${syncBody.skippedCount} already present).`);
+            window.location.reload();
+        } catch (err) {
+            progress.textContent = "";
+            submitBtn.disabled = false;
+            alert("Something went wrong attaching/syncing the S3 bucket.\nReason: " + err.message);
+        }
+    }
+
+    async function detachS3Bucket() {
+        if (!confirm("Are you sure you want to detach the S3 bucket?")) return;
+        try {
+            const res = await fetch(`/api/v2/projects/<%= Admin %>/<%= PName %>/s3-bucket`, {
+                method: 'DELETE',
+            });
+            const data = await res.json();
+            if (res.ok && data.success) {
+                alert("S3 Bucket detached.");
+                window.location.reload();
+            } else {
+                alert("Failed to detach S3 bucket: " + (data.error || res.statusText));
+            }
+        } catch (err) {
+            alert("Error detaching S3 bucket: " + err.message);
+        }
+    }
+
+    document.addEventListener("DOMContentLoaded", () => {
+        loadS3BucketStatus();
+    });
+</script>
diff --git a/views/settings/imagesSettings.ejs b/views/settings/imagesSettings.ejs
index 45d4eb75..331c8429 100644
--- a/views/settings/imagesSettings.ejs
+++ b/views/settings/imagesSettings.ejs
@@ -46,15 +46,19 @@
                         </div>
                     </div>
                     <div class="form-row">
-                        <div class="form-group col-md-4">
+                        <div class="form-group col-md-3">
                             <label for="s3_prefix">Prefix (Optional)</label>
                             <input id="s3_prefix" class="form-control" type="text" placeholder="images/">
                         </div>
-                        <div class="form-group col-md-4">
+                        <div class="form-group col-md-3">
                             <label for="s3_endpoint">S3 Endpoint</label>
                             <input id="s3_endpoint" name="endpoint" aria-label="S3 Endpoint URL" type="text" class="form-control" autocomplete="off">
                         </div>
-                        <div class="form-group col-md-4">
+                        <div class="form-group col-md-3">
+                            <label for="s3_max_images">Max Image Limit (Optional)</label>
+                            <input id="s3_max_images" name="max_images" type="number" min="1" class="form-control" placeholder="Unlimited (e.g. 100)">
+                        </div>
+                        <div class="form-group col-md-3">
                             <label for="s3_access_key_id">Access Key ID</label>
                             <input id="s3_access_key_id" class="form-control" type="text" autocomplete="off">
                         </div>
@@ -136,6 +140,7 @@
 		const accessKeyId = document.getElementById('s3_access_key_id').value;
 		const secretAccessKey = document.getElementById('s3_secret_access_key').value;
         const endpoint = document.getElementById("s3_endpoint").value;
+        const maxImages = document.getElementById("s3_max_images").value;
 
 		if (!bucketName || !region) {
 			alert("Bucket Name and Region are required.");
@@ -159,6 +164,7 @@
 					AccessKeyId: accessKeyId,
 					SecretAccessKey: secretAccessKey,
                     Endpoint: endpoint,
+                    MaxImages: maxImages ? parseInt(maxImages, 10) : null,
 				}),
 			});
 
diff --git a/views/training/inceptionSettings.ejs b/views/training/inceptionSettings.ejs
index fada25f2..117dcee1 100644
--- a/views/training/inceptionSettings.ejs
+++ b/views/training/inceptionSettings.ejs
@@ -108,7 +108,7 @@
                                     <div class="form-group col-md-6">
                                         <h5>
                                             <label for="upload-inference">
-                                                Upload inference file (.jpg, .png, .gif, .tif, .mp4)
+                                                Upload inference file (.jpg, .png, .gif, .tif, .mp4, .mov, .zip)
                                             </label>
                                         </h5>
 
@@ -117,8 +117,10 @@
                                             name="upload-inference"
                                             type="file"
                                             class="form-control-file"
+                                            accept=".jpg,.jpeg,.png,.gif,.tif,.tiff,.mp4,.mov,.zip"
                                             required
                                         >
+                                        <p class="text-muted small">Upload single images/videos or a .zip archive containing images/videos.</p>
                                     </div>
                                 </div>
 
@@ -134,7 +136,9 @@
                                 </div>
                             </form>
 
-                            <br><br>
+                            <br>
+                            <%- include('../includes/s3BucketForm') %>
+                            <br>
 
                             <!-- Inception Inference Config -->
                             <form
diff --git a/views/training/megadetectorSettings.ejs b/views/training/megadetectorSettings.ejs
index b2f7f09d..97ff30d4 100644
--- a/views/training/megadetectorSettings.ejs
+++ b/views/training/megadetectorSettings.ejs
@@ -27,9 +27,10 @@
                                         <div class="form-row">
                                             <div class="form-group col-md-6">
                                                 <h5><label for="upload-md-inference" style="width: 200%;">Upload
-                                                        inference file (.jpg, .png, .gif, .tif, .mp4)</label></h5>
+                                                        inference file (.jpg, .png, .gif, .tif, .mp4, .mov, .zip)</label></h5>
                                                 <input id="upload-md-inference" name="upload-inference" type="file"
-                                                    class="form-control-file" required>
+                                                    class="form-control-file" accept=".jpg,.jpeg,.png,.gif,.tif,.tiff,.mp4,.mov,.zip" required>
+                                                <p>Upload single images/videos or a .zip archive containing images/videos.</p>
                                             </div>
                                         </div>
                                         <div class="col text-left">
@@ -38,6 +39,7 @@
                                         </div>
                                     </form>
                                     <br>
+                                    <%- include('../includes/s3BucketForm') %>
                                     <br>
                                     <form class="Configure Project For MegaDetector" action="/megadetector-inf"
                                         enctype="multipart/form-data" id="megadetector-inf" method="post">
diff --git a/views/training/yolovXInferenceSettings.ejs b/views/training/yolovXInferenceSettings.ejs
index 7d1ce7f4..2fbfb599 100644
--- a/views/training/yolovXInferenceSettings.ejs
+++ b/views/training/yolovXInferenceSettings.ejs
@@ -84,10 +84,10 @@
                                         <div class="form-row">
                                             <div class="form-group col-md-6">
                                                 <h5><label for="upload-inference" style="width: 200%;">Upload inference
-                                                        file (.jpg, .png, .gif, .tif, .mp4)</label></h5>
+                                                        file (.jpg, .png, .gif, .tif, .mp4, .mov, .zip)</label></h5>
                                                 <input id="upload-inference" name="upload-inference" type="file"
-                                                    class="form-control-file" required>
-                                                <p>This is for doing predict or track for the YOLO MODE.</p>
+                                                    class="form-control-file" accept=".jpg,.jpeg,.png,.gif,.tif,.tiff,.mp4,.mov,.zip" required>
+                                                <p>This is for doing predict or track for the YOLO MODE. You can also upload a .zip archive containing images or videos.</p>
                                             </div>
                                         </div>
                                         <div class="col text-left">
@@ -96,6 +96,7 @@
                                         </div>
                                     </form>
                                     <br>
+                                    <%- include('../includes/s3BucketForm') %>
                                     <br>
                                     <form class="Configure Project For YOLO" action="/yolo-inf"
                                         enctype="multipart/form-data" id="yolo-inf" method=post>

From f47f43730ce311c6ba2eba909de9093aa6318ac3 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:14:42 -0700
Subject: [PATCH 08/18] Finalize inference dataset option resolution and test
 suite alignment

Co-authored-by: multica-agent <github@multica.ai>
---
 routes/inference/uploadInferenceFile.js          |  4 +---
 runs/summary.json                                | 16 ++++++++--------
 .../integration/inferenceDatasetPipeline.test.js |  1 +
 tests/run_summary.md                             |  2 +-
 tests/summary.json                               | 16 ++++++++--------
 5 files changed, 19 insertions(+), 20 deletions(-)

diff --git a/routes/inference/uploadInferenceFile.js b/routes/inference/uploadInferenceFile.js
index af1bebfc..0548a8a3 100644
--- a/routes/inference/uploadInferenceFile.js
+++ b/routes/inference/uploadInferenceFile.js
@@ -25,7 +25,7 @@ async function uploadInferenceFile(req, res) {
     if (!validFileNames.includes(ext)) {
         res.send({
             Success:
-                "ERROR: Wrong filetype. Must be type .png, .jpg, .jpeg, .tif, .gif, .mp4, .mov, .zip, or .7z",
+                "ERROR: Wrong filetype. Must be type .png, .jpg, .jpeg, .tif, .gif, .mp4, .mov, or .zip",
         });
     } else {
         if (!fs.existsSync(inferencePath)) {
@@ -68,8 +68,6 @@ async function uploadInferenceFile(req, res) {
 
         res.send({
             Success: "Your inference file has been uploaded and saved",
-            filename: inferenceFile.name,
-            extractedPath,
         });
     }
 }
diff --git a/runs/summary.json b/runs/summary.json
index 48901e31..30f67eca 100644
--- a/runs/summary.json
+++ b/runs/summary.json
@@ -1,10 +1,10 @@
 {
   "runName": "all_runs_summary",
-  "runDir": "/home/fc/code/njobvu-ai/runs",
+  "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs",
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": null,
-  "generatedAt": "2026-08-17T21:26:41.634Z",
+  "generatedAt": "2026-08-18T23:14:36.646Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -24,10 +24,10 @@
   "runs": [
     {
       "runName": "detect_summary_run",
-      "runDir": "/home/fc/code/njobvu-ai/runs/detect/train/detect_summary_run",
+      "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs/detect/train/detect_summary_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-17T21:26:41.633Z",
+      "generatedAt": "2026-08-18T23:14:36.646Z",
       "artifactCount": 1,
       "imageCount": 0,
       "config": {
@@ -47,10 +47,10 @@
       "visualPlots": [],
       "findings": [],
       "recommendations": [],
-      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-17T21:26:41.633Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
-      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-17T21:26:41.633Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
+      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
+      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-17T21:26:41.634Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
-  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-17T21:26:41.634Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
+  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
+  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
 }
\ No newline at end of file
diff --git a/tests/integration/inferenceDatasetPipeline.test.js b/tests/integration/inferenceDatasetPipeline.test.js
index 2f724280..66e27418 100644
--- a/tests/integration/inferenceDatasetPipeline.test.js
+++ b/tests/integration/inferenceDatasetPipeline.test.js
@@ -172,6 +172,7 @@ describe("Inference Dataset Pipeline & S3 Max Image Limit", () => {
                 "",
                 undefined,
                 undefined,
+                "",
                 50,
             );
         });
diff --git a/tests/run_summary.md b/tests/run_summary.md
index f852332a..b78335d2 100644
--- a/tests/run_summary.md
+++ b/tests/run_summary.md
@@ -2,7 +2,7 @@
 
 - **Type**: AGGREGATED ALL-RUNS REPORT
 - **Project**: tmp_test_run
-- **Generated At**: 2026-08-17T21:25:53.329Z
+- **Generated At**: 2026-08-18T23:14:35.695Z
 - **Total Runs Analyzed**: 1 (1 training, 0 inference)
 
 ## Executive Summary & Findings
diff --git a/tests/summary.json b/tests/summary.json
index a650f47a..fd1988b2 100644
--- a/tests/summary.json
+++ b/tests/summary.json
@@ -1,10 +1,10 @@
 {
   "runName": "tmp_test_run_all_runs_summary",
-  "runDir": "/home/fc/code/njobvu-ai/tests",
+  "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests",
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": "tmp_test_run",
-  "generatedAt": "2026-08-17T21:25:53.329Z",
+  "generatedAt": "2026-08-18T23:14:35.695Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -25,10 +25,10 @@
   "runs": [
     {
       "runName": "tmp_test_run",
-      "runDir": "/home/fc/code/njobvu-ai/tests/tmp_test_run",
+      "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests/tmp_test_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-17T21:25:53.324Z",
+      "generatedAt": "2026-08-18T23:14:35.695Z",
       "artifactCount": 4,
       "imageCount": 1,
       "config": {
@@ -69,10 +69,10 @@
       "recommendations": [
         "Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment."
       ],
-      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-17T21:25:53.324Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
-      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-17T21:25:53.324Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
+      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
+      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-17T21:25:53.329Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
-  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-17T21:25:53.329Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
+  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
+  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
 }
\ No newline at end of file

From e4f309142d4e9b4b5b712ffb3b8550dcb8071b34 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:29:24 -0700
Subject: [PATCH 09/18] refactor(s3): remove MaxImages DB column and handle
 maxImages per-sync/inference run request payload

Co-authored-by: multica-agent <github@multica.ai>
---
 db/migrations.sql                             |  3 +-
 queries/s3/s3.js                              | 63 +++++--------------
 routes/api/v2/s3Buckets.js                    | 11 +---
 runs/summary.json                             | 12 ++--
 .../inferenceDatasetPipeline.test.js          |  6 +-
 tests/integration/s3Buckets.test.js           |  5 +-
 tests/run_summary.md                          |  2 +-
 tests/summary.json                            | 12 ++--
 views/includes/s3BucketForm.ejs               |  5 +-
 9 files changed, 40 insertions(+), 79 deletions(-)

diff --git a/db/migrations.sql b/db/migrations.sql
index 2b77f987..864395ac 100644
--- a/db/migrations.sql
+++ b/db/migrations.sql
@@ -2,5 +2,4 @@ CREATE TABLE Users (Username TEXT NOT NULL PRIMARY KEY, Password TEXT NOT NULL,
 INSERT INTO Users (Username, Password, FirstName, LastName, Email) VALUES ('ZeroUser', 'Temp123', 'ZeroUser', 'ZeroUser', 'ZeroUser');
 CREATE TABLE Projects (PName VARCHAR NOT NULL, PDescription VARCHAR NOT NULL, AutoSave INTEGER NOT NULL DEFAULT 0, Admin TEXT NOT NULL DEFAULT 'ZeroUser', Validate VARCHAR NOT NULL DEFAULT 0, FOREIGN KEY(Admin) REFERENCES Users(Username), PRIMARY KEY(PName, Admin));
 CREATE TABLE Access (Username TEXT NOT NULL, PName VARCHAR NOT NULL,  Admin TEXT NOT NULL, FOREIGN KEY(Username) REFERENCES Users(Username), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
-CREATE TABLE S3Buckets (PName VARCHAR NOT NULL, Admin TEXT NOT NULL, BucketName VARCHAR NOT NULL, Region VARCHAR NOT NULL, Prefix VARCHAR NOT NULL DEFAULT '', AccessKeyId VARCHAR, SecretAccessKey VARCHAR, LastSyncedAt TEXT, Endpoint TEXT, MaxImages INTEGER DEFAULT NULL, PRIMARY KEY(PName, Admin), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
-ALTER TABLE S3Buckets ADD COLUMN MaxImages INTEGER DEFAULT NULL;
+CREATE TABLE S3Buckets (PName VARCHAR NOT NULL, Admin TEXT NOT NULL, BucketName VARCHAR NOT NULL, Region VARCHAR NOT NULL, Prefix VARCHAR NOT NULL DEFAULT '', AccessKeyId VARCHAR, SecretAccessKey VARCHAR, LastSyncedAt TEXT, Endpoint TEXT, PRIMARY KEY(PName, Admin), FOREIGN KEY(PName) REFERENCES Projects(PName), FOREIGN KEY(Admin) REFERENCES Projects(Admin));
diff --git a/queries/s3/s3.js b/queries/s3/s3.js
index 6331da1b..e0ab0346 100644
--- a/queries/s3/s3.js
+++ b/queries/s3/s3.js
@@ -8,54 +8,25 @@ module.exports = {
             prefix,
             accessKeyId,
             secretAccessKey,
-            endpoint,
-            maxImages
+            endpoint
         ) {
-            const parsedMax =
-                maxImages !== undefined && maxImages !== null && maxImages !== ""
-                    ? parseInt(maxImages, 10)
-                    : null;
-            const validMax = Number.isFinite(parsedMax) && parsedMax > 0 ? parsedMax : null;
-
-            try {
-                const queryWithMax =
-                    "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint, MaxImages) " +
-                    "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?) " +
-                    "ON CONFLICT(PName, Admin) DO UPDATE SET " +
-                    "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
-                    "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint, " +
-                    "MaxImages = excluded.MaxImages";
-
-                return await global.managedDbClient.run(queryWithMax, [
-                    projectName,
-                    admin,
-                    bucketName,
-                    region,
-                    prefix || "",
-                    accessKeyId || null,
-                    secretAccessKey || null,
-                    endpoint || "",
-                    validMax,
-                ]);
-            } catch (err) {
-                const queryBase =
-                    "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint) " +
-                    "VALUES (?, ?, ?, ?, ?, ?, ?, ?) " +
-                    "ON CONFLICT(PName, Admin) DO UPDATE SET " +
-                    "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
-                    "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint";
+            const query =
+                "INSERT INTO S3Buckets (PName, Admin, BucketName, Region, Prefix, AccessKeyId, SecretAccessKey, Endpoint) " +
+                "VALUES (?, ?, ?, ?, ?, ?, ?, ?) " +
+                "ON CONFLICT(PName, Admin) DO UPDATE SET " +
+                "BucketName = excluded.BucketName, Region = excluded.Region, Prefix = excluded.Prefix, " +
+                "AccessKeyId = excluded.AccessKeyId, SecretAccessKey = excluded.SecretAccessKey, Endpoint = excluded.Endpoint";
 
-                return await global.managedDbClient.run(queryBase, [
-                    projectName,
-                    admin,
-                    bucketName,
-                    region,
-                    prefix || "",
-                    accessKeyId || null,
-                    secretAccessKey || null,
-                    endpoint || "",
-                ]);
-            }
+            return await global.managedDbClient.run(query, [
+                projectName,
+                admin,
+                bucketName,
+                region,
+                prefix || "",
+                accessKeyId || null,
+                secretAccessKey || null,
+                endpoint || "",
+            ]);
         },
         getBucket: async function(projectName, admin) {
             const query =
diff --git a/routes/api/v2/s3Buckets.js b/routes/api/v2/s3Buckets.js
index 0dc6ebb7..6fa220a5 100644
--- a/routes/api/v2/s3Buckets.js
+++ b/routes/api/v2/s3Buckets.js
@@ -38,8 +38,6 @@ async function attachS3Bucket(req, res) {
     const AccessKeyId = trimOrUndefined(body.AccessKeyId);
     const SecretAccessKey = trimOrUndefined(body.SecretAccessKey);
     const Endpoint = trimOrUndefined(body.Endpoint);
-    const rawMaxImages = body.MaxImages !== undefined ? body.MaxImages : (body.maxImages !== undefined ? body.maxImages : body.max_images);
-    const MaxImages = (rawMaxImages !== undefined && rawMaxImages !== null && rawMaxImages !== "") ? parseInt(rawMaxImages, 10) : null;
 
     if (!isOwner(req, admin)) {
         return res.status(403).json({ success: false, error: "Not authorized for this project" });
@@ -72,8 +70,7 @@ async function attachS3Bucket(req, res) {
             Prefix || "",
             AccessKeyId,
             SecretAccessKey,
-            Endpoint || "",
-            MaxImages
+            Endpoint || ""
         );
 
         return res.status(200).json({ success: true });
@@ -105,8 +102,6 @@ async function getS3Bucket(req, res) {
             return res.status(404).json({ success: false, error: "No S3 bucket attached" });
         }
 
-        const maxImg = row.MaxImages != null ? Number(row.MaxImages) : null;
-
         return res.status(200).json({
             success: true,
             bucket: {
@@ -114,8 +109,6 @@ async function getS3Bucket(req, res) {
                 Region: row.Region,
                 Prefix: row.Prefix,
                 Endpoint: row.Endpoint,
-                MaxImages: maxImg,
-                max_images: maxImg,
                 LastSyncedAt: row.LastSyncedAt,
                 hasCredentials: !!row.AccessKeyId,
             },
@@ -166,7 +159,7 @@ async function syncS3Bucket(req, res) {
 
         const rawMaxLimit = (req.body && (req.body.maxImages || req.body.MaxImages || req.body.max_images || req.body.limit)) || (req.query && req.query.limit);
         const parsedMaxLimit = rawMaxLimit !== undefined && rawMaxLimit !== null && rawMaxLimit !== "" ? parseInt(rawMaxLimit, 10) : null;
-        const maxImages = parsedMaxLimit !== null ? parsedMaxLimit : (bucket.MaxImages != null ? Number(bucket.MaxImages) : null);
+        const maxImages = parsedMaxLimit !== null ? parsedMaxLimit : null;
 
         const s3Client = buildS3Client({
             region: bucket.Region,
diff --git a/runs/summary.json b/runs/summary.json
index 30f67eca..9624ae45 100644
--- a/runs/summary.json
+++ b/runs/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": null,
-  "generatedAt": "2026-08-18T23:14:36.646Z",
+  "generatedAt": "2026-08-18T23:29:21.412Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -27,7 +27,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs/detect/train/detect_summary_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:14:36.646Z",
+      "generatedAt": "2026-08-18T23:29:21.412Z",
       "artifactCount": 1,
       "imageCount": 0,
       "config": {
@@ -47,10 +47,10 @@
       "visualPlots": [],
       "findings": [],
       "recommendations": [],
-      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
-      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
+      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
+      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
-  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:14:36.646Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
+  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
+  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
 }
\ No newline at end of file
diff --git a/tests/integration/inferenceDatasetPipeline.test.js b/tests/integration/inferenceDatasetPipeline.test.js
index 66e27418..f660f155 100644
--- a/tests/integration/inferenceDatasetPipeline.test.js
+++ b/tests/integration/inferenceDatasetPipeline.test.js
@@ -148,8 +148,8 @@ describe("Inference Dataset Pipeline & S3 Max Image Limit", () => {
         });
     });
 
-    describe("attachS3Bucket with MaxImages field", () => {
-        it("saves MaxImages when configuring bucket", async () => {
+    describe("attachS3Bucket without MaxImages column", () => {
+        it("attaches bucket successfully without MaxImages in DB", async () => {
             const projectDir = path.join(tmpDir, "public/projects/testuser-test-project");
             fs.mkdirSync(projectDir, { recursive: true });
 
@@ -160,7 +160,6 @@ describe("Inference Dataset Pipeline & S3 Max Image Limit", () => {
                 .send({
                     BucketName: "test-bucket",
                     Region: "us-east-1",
-                    MaxImages: 50,
                 });
 
             expect(res.statusCode).toBe(200);
@@ -173,7 +172,6 @@ describe("Inference Dataset Pipeline & S3 Max Image Limit", () => {
                 undefined,
                 undefined,
                 "",
-                50,
             );
         });
     });
diff --git a/tests/integration/s3Buckets.test.js b/tests/integration/s3Buckets.test.js
index d8c0bd0d..e48f551a 100644
--- a/tests/integration/s3Buckets.test.js
+++ b/tests/integration/s3Buckets.test.js
@@ -97,17 +97,15 @@ describe('S3 Bucket Routes', () => {
                 'AKIA...',
                 'secret',
                 '',
-                null,
             );
         });
 
-        it('attaches a bucket with MaxImages limit', async () => {
+        it('attaches a bucket successfully', async () => {
             const res = await request(app)
                 .post('/api/v2/projects/testuser/test-project/s3-bucket')
                 .set('Cookie', ['Username=testuser'])
                 .send({
                     ...validBody,
-                    MaxImages: 50,
                 });
 
             expect(res.statusCode).toBe(200);
@@ -121,7 +119,6 @@ describe('S3 Bucket Routes', () => {
                 'AKIA...',
                 'secret',
                 '',
-                50,
             );
         });
 
diff --git a/tests/run_summary.md b/tests/run_summary.md
index b78335d2..19dab678 100644
--- a/tests/run_summary.md
+++ b/tests/run_summary.md
@@ -2,7 +2,7 @@
 
 - **Type**: AGGREGATED ALL-RUNS REPORT
 - **Project**: tmp_test_run
-- **Generated At**: 2026-08-18T23:14:35.695Z
+- **Generated At**: 2026-08-18T23:29:21.095Z
 - **Total Runs Analyzed**: 1 (1 training, 0 inference)
 
 ## Executive Summary & Findings
diff --git a/tests/summary.json b/tests/summary.json
index fd1988b2..b66da198 100644
--- a/tests/summary.json
+++ b/tests/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": "tmp_test_run",
-  "generatedAt": "2026-08-18T23:14:35.695Z",
+  "generatedAt": "2026-08-18T23:29:21.095Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -28,7 +28,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests/tmp_test_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:14:35.695Z",
+      "generatedAt": "2026-08-18T23:29:21.095Z",
       "artifactCount": 4,
       "imageCount": 1,
       "config": {
@@ -69,10 +69,10 @@
       "recommendations": [
         "Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment."
       ],
-      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
-      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
+      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
+      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
-  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:14:35.695Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
+  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
+  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
 }
\ No newline at end of file
diff --git a/views/includes/s3BucketForm.ejs b/views/includes/s3BucketForm.ejs
index e703111d..d2f5702d 100644
--- a/views/includes/s3BucketForm.ejs
+++ b/views/includes/s3BucketForm.ejs
@@ -117,7 +117,6 @@
                     AccessKeyId: accessKeyId,
                     SecretAccessKey: secretAccessKey,
                     Endpoint: endpoint,
-                    MaxImages: maxImages ? parseInt(maxImages, 10) : null,
                 }),
             });
 
@@ -130,6 +129,10 @@
 
             const syncRes = await fetch(`/api/v2/projects/<%= Admin %>/<%= PName %>/s3-bucket/sync`, {
                 method: 'POST',
+                headers: { 'Content-Type': 'application/json' },
+                body: JSON.stringify({
+                    maxImages: maxImages ? parseInt(maxImages, 10) : null,
+                }),
             });
 
             const syncBody = await syncRes.json();

From 799a646e184ad317a75caad6e9ea2e8809832ba2 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:35:17 -0700
Subject: [PATCH 10/18] fix(s3): resolve sync loop break condition bug when
 existing images are present

Co-authored-by: multica-agent <github@multica.ai>
---
 routes/api/v2/s3Buckets.js |  2 +-
 runs/summary.json          | 12 ++++++------
 tests/run_summary.md       |  2 +-
 tests/summary.json         | 12 ++++++------
 4 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/routes/api/v2/s3Buckets.js b/routes/api/v2/s3Buckets.js
index 6fa220a5..347cd2d5 100644
--- a/routes/api/v2/s3Buckets.js
+++ b/routes/api/v2/s3Buckets.js
@@ -175,7 +175,7 @@ async function syncS3Bucket(req, res) {
         let skippedCount = 0;
 
         for (const key of objectKeys) {
-            if (Number.isFinite(maxImages) && maxImages > 0 && (syncedImages.length >= maxImages || existingImages.size >= maxImages)) {
+            if (Number.isFinite(maxImages) && maxImages > 0 && syncedImages.length >= maxImages) {
                 break;
             }
 
diff --git a/runs/summary.json b/runs/summary.json
index 9624ae45..dcff6214 100644
--- a/runs/summary.json
+++ b/runs/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": null,
-  "generatedAt": "2026-08-18T23:29:21.412Z",
+  "generatedAt": "2026-08-18T23:35:12.022Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -27,7 +27,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs/detect/train/detect_summary_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:29:21.412Z",
+      "generatedAt": "2026-08-18T23:35:12.021Z",
       "artifactCount": 1,
       "imageCount": 0,
       "config": {
@@ -47,10 +47,10 @@
       "visualPlots": [],
       "findings": [],
       "recommendations": [],
-      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
-      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
+      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:12.021Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
+      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:12.021Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
-  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:29:21.412Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
+  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:35:12.022Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
+  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:35:12.022Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
 }
\ No newline at end of file
diff --git a/tests/run_summary.md b/tests/run_summary.md
index 19dab678..1a9b0787 100644
--- a/tests/run_summary.md
+++ b/tests/run_summary.md
@@ -2,7 +2,7 @@
 
 - **Type**: AGGREGATED ALL-RUNS REPORT
 - **Project**: tmp_test_run
-- **Generated At**: 2026-08-18T23:29:21.095Z
+- **Generated At**: 2026-08-18T23:35:11.278Z
 - **Total Runs Analyzed**: 1 (1 training, 0 inference)
 
 ## Executive Summary & Findings
diff --git a/tests/summary.json b/tests/summary.json
index b66da198..ca0f8ae8 100644
--- a/tests/summary.json
+++ b/tests/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": "tmp_test_run",
-  "generatedAt": "2026-08-18T23:29:21.095Z",
+  "generatedAt": "2026-08-18T23:35:11.278Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -28,7 +28,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests/tmp_test_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:29:21.095Z",
+      "generatedAt": "2026-08-18T23:35:11.274Z",
       "artifactCount": 4,
       "imageCount": 1,
       "config": {
@@ -69,10 +69,10 @@
       "recommendations": [
         "Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment."
       ],
-      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
-      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
+      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:11.274Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
+      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:11.274Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
-  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:29:21.095Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
+  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:35:11.278Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
+  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:35:11.278Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
 }
\ No newline at end of file

From e3ec7644e2efb7bd2fd0e19b642a9872c70ce3be Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:37:17 -0700
Subject: [PATCH 11/18] feat(s3): add JIT streaming default caps and maxKeys
 pagination throttle safeguards

Co-authored-by: multica-agent <github@multica.ai>
---
 runs/summary.json                 | 12 ++++++------
 tests/run_summary.md              |  2 +-
 tests/summary.json                | 12 ++++++------
 utils/inferenceDatasetPipeline.js | 16 +++++++++++-----
 utils/s3Client.js                 |  5 ++++-
 5 files changed, 28 insertions(+), 19 deletions(-)

diff --git a/runs/summary.json b/runs/summary.json
index dcff6214..dc07f09f 100644
--- a/runs/summary.json
+++ b/runs/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": null,
-  "generatedAt": "2026-08-18T23:35:12.022Z",
+  "generatedAt": "2026-08-18T23:37:11.740Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -27,7 +27,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs/detect/train/detect_summary_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:35:12.021Z",
+      "generatedAt": "2026-08-18T23:37:11.740Z",
       "artifactCount": 1,
       "imageCount": 0,
       "config": {
@@ -47,10 +47,10 @@
       "visualPlots": [],
       "findings": [],
       "recommendations": [],
-      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:12.021Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
-      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:12.021Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
+      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
+      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:35:12.022Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
-  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:35:12.022Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
+  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
+  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
 }
\ No newline at end of file
diff --git a/tests/run_summary.md b/tests/run_summary.md
index 1a9b0787..817095a0 100644
--- a/tests/run_summary.md
+++ b/tests/run_summary.md
@@ -2,7 +2,7 @@
 
 - **Type**: AGGREGATED ALL-RUNS REPORT
 - **Project**: tmp_test_run
-- **Generated At**: 2026-08-18T23:35:11.278Z
+- **Generated At**: 2026-08-18T23:37:11.064Z
 - **Total Runs Analyzed**: 1 (1 training, 0 inference)
 
 ## Executive Summary & Findings
diff --git a/tests/summary.json b/tests/summary.json
index ca0f8ae8..5d25eeb5 100644
--- a/tests/summary.json
+++ b/tests/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": "tmp_test_run",
-  "generatedAt": "2026-08-18T23:35:11.278Z",
+  "generatedAt": "2026-08-18T23:37:11.064Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -28,7 +28,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests/tmp_test_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:35:11.274Z",
+      "generatedAt": "2026-08-18T23:37:11.064Z",
       "artifactCount": 4,
       "imageCount": 1,
       "config": {
@@ -69,10 +69,10 @@
       "recommendations": [
         "Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment."
       ],
-      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:11.274Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
-      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:35:11.274Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
+      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
+      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:35:11.278Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
-  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:35:11.278Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
+  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
+  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
 }
\ No newline at end of file
diff --git a/utils/inferenceDatasetPipeline.js b/utils/inferenceDatasetPipeline.js
index daef23fe..655520bd 100644
--- a/utils/inferenceDatasetPipeline.js
+++ b/utils/inferenceDatasetPipeline.js
@@ -4,6 +4,9 @@ const unzipFile = require("./unzipFile");
 const queries = require("../queries/queries");
 const { buildS3Client, listImageObjects, downloadObjectToFile } = require("./s3Client");
 
+const DEFAULT_MAX_IMAGES = 100;
+const SAFETY_CEILING_MAX_IMAGES = 5000;
+
 async function prepareInferenceDataset(options) {
     const {
         PName,
@@ -17,11 +20,15 @@ async function prepareInferenceDataset(options) {
         inferenceUploadPath,
     } = options;
 
-    const effectiveMaxImages = max_images || maxImages || limit;
+    const rawLimit = max_images || maxImages || limit;
+    const parsedLimit = rawLimit ? parseInt(rawLimit, 10) : null;
+    const maxLimit = parsedLimit && Number.isFinite(parsedLimit) && parsedLimit > 0
+        ? Math.min(parsedLimit, SAFETY_CEILING_MAX_IMAGES)
+        : DEFAULT_MAX_IMAGES;
 
     let targetFilePath = inference_file;
 
-    // S3 Bucket stream mode
+    // S3 Bucket stream mode (JIT fetch right before inference run execution)
     if (use_s3_bucket || inference_file === "s3" || inference_file === "s3_bucket") {
         const bucketResult = await queries.managed.getBucket(PName, Admin);
         const bucket = bucketResult && bucketResult.row;
@@ -30,8 +37,6 @@ async function prepareInferenceDataset(options) {
             throw new Error("No S3 bucket attached to this project");
         }
 
-        const rawLimit = effectiveMaxImages || bucket.MaxImages || 100;
-        const maxLimit = parseInt(rawLimit, 10);
         const dateStamp = Date.now();
         const s3StreamFolder = path.join(inferenceUploadPath, `s3_stream_${dateStamp}`);
 
@@ -46,7 +51,8 @@ async function prepareInferenceDataset(options) {
             endpoint: bucket.Endpoint,
         });
 
-        const objectKeys = await listImageObjects(s3Client, bucket.BucketName, bucket.Prefix);
+        // Pass maxLimit directly to listImageObjects so S3 listing stops early
+        const objectKeys = await listImageObjects(s3Client, bucket.BucketName, bucket.Prefix, maxLimit);
         let downloaded = 0;
 
         for (const key of objectKeys) {
diff --git a/utils/s3Client.js b/utils/s3Client.js
index 93e90951..0a46a186 100644
--- a/utils/s3Client.js
+++ b/utils/s3Client.js
@@ -86,7 +86,7 @@ async function verifyBucketAccess(s3Client, bucketName) {
     );
 }
 
-async function listImageObjects(s3Client, bucketName, prefix = "") {
+async function listImageObjects(s3Client, bucketName, prefix = "", maxKeys = null) {
     const objects = [];
     let continuationToken;
 
@@ -106,6 +106,9 @@ async function listImageObjects(s3Client, bucketName, prefix = "") {
 
             if (IMAGE_EXTENSIONS.has(path.extname(object.Key).toLowerCase())) {
                 objects.push(object.Key);
+                if (maxKeys && objects.length >= maxKeys) {
+                    return objects;
+                }
             }
         }
 

From e2ba4f9b4c13d5a4e56ceada35f837ec68fb9508 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:44:58 -0700
Subject: [PATCH 12/18] feat(s3): integrate feature/image-source-column
 streaming changes and sync mode selection into inference pipeline

Co-authored-by: multica-agent <github@multica.ai>
---
 runs/summary.json                             | 12 ++++++------
 .../inferenceDatasetPipeline.test.js          |  2 ++
 tests/integration/s3Buckets.test.js           |  9 ---------
 tests/run_summary.md                          |  2 +-
 tests/summary.json                            | 12 ++++++------
 views/includes/s3BucketForm.ejs               | 19 +++++++++++++++----
 6 files changed, 30 insertions(+), 26 deletions(-)

diff --git a/runs/summary.json b/runs/summary.json
index dc07f09f..3ccaba8c 100644
--- a/runs/summary.json
+++ b/runs/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": null,
-  "generatedAt": "2026-08-18T23:37:11.740Z",
+  "generatedAt": "2026-08-18T23:44:51.720Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -27,7 +27,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs/detect/train/detect_summary_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:37:11.740Z",
+      "generatedAt": "2026-08-18T23:44:51.719Z",
       "artifactCount": 1,
       "imageCount": 0,
       "config": {
@@ -47,10 +47,10 @@
       "visualPlots": [],
       "findings": [],
       "recommendations": [],
-      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
-      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
+      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:51.719Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
+      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:51.719Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
-  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:37:11.740Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
+  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:44:51.720Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
+  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:44:51.720Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
 }
\ No newline at end of file
diff --git a/tests/integration/inferenceDatasetPipeline.test.js b/tests/integration/inferenceDatasetPipeline.test.js
index f660f155..7c3d9a2b 100644
--- a/tests/integration/inferenceDatasetPipeline.test.js
+++ b/tests/integration/inferenceDatasetPipeline.test.js
@@ -18,6 +18,7 @@ jest.mock("../../queries/queries", () => ({
     },
     project: {
         addImages: jest.fn().mockResolvedValue({ success: true }),
+        getAllImages: jest.fn().mockResolvedValue({ success: true, rows: [] }),
     },
 }));
 
@@ -172,6 +173,7 @@ describe("Inference Dataset Pipeline & S3 Max Image Limit", () => {
                 undefined,
                 undefined,
                 "",
+                "download",
             );
         });
     });
diff --git a/tests/integration/s3Buckets.test.js b/tests/integration/s3Buckets.test.js
index c5cad963..6ca28e93 100644
--- a/tests/integration/s3Buckets.test.js
+++ b/tests/integration/s3Buckets.test.js
@@ -104,15 +104,6 @@ describe('S3 Bucket Routes', () => {
                 { fakeClient: true },
                 'my-bucket',
             );
-            expect(queries.managed.attachBucket).toHaveBeenCalledWith(
-                'test-project',
-                'testuser',
-                'my-bucket',
-                'us-east-1',
-                'images/',
-                'AKIA...',
-                'secret',
-                '',
             expect(queries.managed.attachBucket).toHaveBeenCalledWith(
                 'test-project',
                 'testuser',
diff --git a/tests/run_summary.md b/tests/run_summary.md
index 817095a0..9087b268 100644
--- a/tests/run_summary.md
+++ b/tests/run_summary.md
@@ -2,7 +2,7 @@
 
 - **Type**: AGGREGATED ALL-RUNS REPORT
 - **Project**: tmp_test_run
-- **Generated At**: 2026-08-18T23:37:11.064Z
+- **Generated At**: 2026-08-18T23:44:50.477Z
 - **Total Runs Analyzed**: 1 (1 training, 0 inference)
 
 ## Executive Summary & Findings
diff --git a/tests/summary.json b/tests/summary.json
index 5d25eeb5..c98603d8 100644
--- a/tests/summary.json
+++ b/tests/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": "tmp_test_run",
-  "generatedAt": "2026-08-18T23:37:11.064Z",
+  "generatedAt": "2026-08-18T23:44:50.477Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -28,7 +28,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests/tmp_test_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:37:11.064Z",
+      "generatedAt": "2026-08-18T23:44:50.477Z",
       "artifactCount": 4,
       "imageCount": 1,
       "config": {
@@ -69,10 +69,10 @@
       "recommendations": [
         "Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment."
       ],
-      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
-      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
+      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
+      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
-  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:37:11.064Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
+  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
+  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
 }
\ No newline at end of file
diff --git a/views/includes/s3BucketForm.ejs b/views/includes/s3BucketForm.ejs
index d2f5702d..7f01f496 100644
--- a/views/includes/s3BucketForm.ejs
+++ b/views/includes/s3BucketForm.ejs
@@ -23,15 +23,22 @@
                     </div>
                 </div>
                 <div class="form-row">
-                    <div class="form-group col-md-4">
+                    <div class="form-group col-md-3">
                         <label for="s3_prefix">Prefix (Optional)</label>
                         <input id="s3_prefix" class="form-control" type="text" placeholder="images/">
                     </div>
-                    <div class="form-group col-md-4">
+                    <div class="form-group col-md-3">
                         <label for="s3_endpoint">S3 Endpoint (Optional)</label>
                         <input id="s3_endpoint" type="text" class="form-control" placeholder="https://s3.amazonaws.com">
                     </div>
-                    <div class="form-group col-md-4">
+                    <div class="form-group col-md-3">
+                        <label for="s3_sync_mode">Sync Mode</label>
+                        <select id="s3_sync_mode" class="form-control">
+                            <option value="stream" selected>Stream on demand</option>
+                            <option value="download">Download all</option>
+                        </select>
+                    </div>
+                    <div class="form-group col-md-3">
                         <label for="s3_max_images">Max Image Limit (Optional)</label>
                         <input id="s3_max_images" name="max_images" type="number" min="1" class="form-control" placeholder="Unlimited (e.g. 100)">
                     </div>
@@ -68,12 +75,14 @@
                     if (statusEl) {
                         statusEl.classList.remove('d-none');
                         let limitText = b.MaxImages ? ` (Max Images: ${b.MaxImages})` : ' (Unlimited images)';
-                        statusEl.innerHTML = `<strong>Attached S3 Bucket:</strong> <code>${b.BucketName}</code> [Region: <code>${b.Region}</code>]${limitText}`;
+                        let modeText = b.SyncMode ? ` [Mode: ${b.SyncMode}]` : '';
+                        statusEl.innerHTML = `<strong>Attached S3 Bucket:</strong> <code>${b.BucketName}</code> [Region: <code>${b.Region}</code>]${modeText}${limitText}`;
                     }
                     if (document.getElementById('s3_bucket_name')) document.getElementById('s3_bucket_name').value = b.BucketName || '';
                     if (document.getElementById('s3_region')) document.getElementById('s3_region').value = b.Region || '';
                     if (document.getElementById('s3_prefix')) document.getElementById('s3_prefix').value = b.Prefix || '';
                     if (document.getElementById('s3_endpoint')) document.getElementById('s3_endpoint').value = b.Endpoint || '';
+                    if (document.getElementById('s3_sync_mode')) document.getElementById('s3_sync_mode').value = b.SyncMode || 'stream';
                     if (b.MaxImages && document.getElementById('s3_max_images')) {
                         document.getElementById('s3_max_images').value = b.MaxImages;
                     }
@@ -93,6 +102,7 @@
         const accessKeyId = document.getElementById('s3_access_key_id').value;
         const secretAccessKey = document.getElementById('s3_secret_access_key').value;
         const endpoint = document.getElementById("s3_endpoint").value;
+        const syncMode = document.getElementById("s3_sync_mode").value;
         const maxImages = document.getElementById("s3_max_images").value;
 
         if (!bucketName || !region) {
@@ -117,6 +127,7 @@
                     AccessKeyId: accessKeyId,
                     SecretAccessKey: secretAccessKey,
                     Endpoint: endpoint,
+                    SyncMode: syncMode,
                 }),
             });
 

From b991916a0f0710932b8c54d9f85af51c021bcb59 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:51:10 -0700
Subject: [PATCH 13/18] feat(inference): add attached S3 bucket option to
 inference settings views and dropdown menus

Co-authored-by: multica-agent <github@multica.ai>
---
 routes/pages/getInceptionSettingsPage.js      |  5 +++
 routes/pages/getMegadetectorSettingsPage.js   |  5 +++
 routes/pages/getYoloXInferenceSettingsPage.js |  5 +++
 runs/summary.json                             | 12 +++----
 tests/run_summary.md                          |  2 +-
 tests/summary.json                            | 12 +++----
 utils/inferenceDatasetPipeline.js             | 34 ++++++++++++++++---
 views/training/inceptionSettings.ejs          |  4 ++-
 views/training/megadetectorSettings.ejs       |  4 +--
 views/training/yolovXInferenceSettings.ejs    |  6 ++--
 10 files changed, 66 insertions(+), 23 deletions(-)

diff --git a/routes/pages/getInceptionSettingsPage.js b/routes/pages/getInceptionSettingsPage.js
index 0690cda2..b37d0f79 100644
--- a/routes/pages/getInceptionSettingsPage.js
+++ b/routes/pages/getInceptionSettingsPage.js
@@ -131,6 +131,11 @@ async function getYoloXInferencePage(req, res) {
     var global_inference_upload = await readdirAsync(inference_upload_path);
     global_inference_upload.push(project_path + "/images");
 
+    const attachedBucket = await queries.managed.getBucket(PName, admin);
+    if (attachedBucket && attachedBucket.row) {
+        global_inference_upload.push("s3");
+    }
+
     // get runs
     var runs = await readdirAsync(log_path);
     runs = runs.reverse();
diff --git a/routes/pages/getMegadetectorSettingsPage.js b/routes/pages/getMegadetectorSettingsPage.js
index 40aee167..a1bef9d5 100644
--- a/routes/pages/getMegadetectorSettingsPage.js
+++ b/routes/pages/getMegadetectorSettingsPage.js
@@ -56,6 +56,11 @@ async function getMegadetectorSettingsPage(req, res) {
     var global_inference_upload = await readdirAsync(inference_upload_path);
     global_inference_upload.push(project_path + "/images");
 
+    const attachedBucket = await queries.managed.getBucket(PName, admin);
+    if (attachedBucket && attachedBucket.row) {
+        global_inference_upload.push("s3");
+    }
+
     res.render("training/megadetectorSettings", {
         title: "megadetectorSettings",
         user: req.cookies.Username,
diff --git a/routes/pages/getYoloXInferenceSettingsPage.js b/routes/pages/getYoloXInferenceSettingsPage.js
index 7274947a..5e17c568 100644
--- a/routes/pages/getYoloXInferenceSettingsPage.js
+++ b/routes/pages/getYoloXInferenceSettingsPage.js
@@ -131,6 +131,11 @@ async function getYoloXInferencePage(req, res) {
     var global_inference_upload = await readdirAsync(inference_upload_path);
     global_inference_upload.push(project_path + "/images");
 
+    const attachedBucket = await queries.managed.getBucket(PName, admin);
+    if (attachedBucket && attachedBucket.row) {
+        global_inference_upload.push("s3");
+    }
+
     // get runs
     var runs = await readdirAsync(log_path);
     runs = runs.reverse();
diff --git a/runs/summary.json b/runs/summary.json
index 3ccaba8c..b2d2e921 100644
--- a/runs/summary.json
+++ b/runs/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": null,
-  "generatedAt": "2026-08-18T23:44:51.720Z",
+  "generatedAt": "2026-08-18T23:51:05.339Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -27,7 +27,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs/detect/train/detect_summary_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:44:51.719Z",
+      "generatedAt": "2026-08-18T23:51:05.338Z",
       "artifactCount": 1,
       "imageCount": 0,
       "config": {
@@ -47,10 +47,10 @@
       "visualPlots": [],
       "findings": [],
       "recommendations": [],
-      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:51.719Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
-      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:51.719Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
+      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:05.338Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
+      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:05.338Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:44:51.720Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
-  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:44:51.720Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
+  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:51:05.339Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
+  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:51:05.339Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
 }
\ No newline at end of file
diff --git a/tests/run_summary.md b/tests/run_summary.md
index 9087b268..e5cdc69b 100644
--- a/tests/run_summary.md
+++ b/tests/run_summary.md
@@ -2,7 +2,7 @@
 
 - **Type**: AGGREGATED ALL-RUNS REPORT
 - **Project**: tmp_test_run
-- **Generated At**: 2026-08-18T23:44:50.477Z
+- **Generated At**: 2026-08-18T23:51:04.138Z
 - **Total Runs Analyzed**: 1 (1 training, 0 inference)
 
 ## Executive Summary & Findings
diff --git a/tests/summary.json b/tests/summary.json
index c98603d8..c3f57a4d 100644
--- a/tests/summary.json
+++ b/tests/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": "tmp_test_run",
-  "generatedAt": "2026-08-18T23:44:50.477Z",
+  "generatedAt": "2026-08-18T23:51:04.138Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -28,7 +28,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests/tmp_test_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:44:50.477Z",
+      "generatedAt": "2026-08-18T23:51:04.137Z",
       "artifactCount": 4,
       "imageCount": 1,
       "config": {
@@ -69,10 +69,10 @@
       "recommendations": [
         "Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment."
       ],
-      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
-      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
+      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:04.137Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
+      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:04.137Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
-  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:44:50.477Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
+  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:51:04.138Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
+  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:51:04.138Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
 }
\ No newline at end of file
diff --git a/utils/inferenceDatasetPipeline.js b/utils/inferenceDatasetPipeline.js
index 655520bd..396e8074 100644
--- a/utils/inferenceDatasetPipeline.js
+++ b/utils/inferenceDatasetPipeline.js
@@ -28,11 +28,37 @@ async function prepareInferenceDataset(options) {
 
     let targetFilePath = inference_file;
 
-    // S3 Bucket stream mode (JIT fetch right before inference run execution)
-    if (use_s3_bucket || inference_file === "s3" || inference_file === "s3_bucket") {
-        const bucketResult = await queries.managed.getBucket(PName, Admin);
-        const bucket = bucketResult && bucketResult.row;
+    const isS3Selection = use_s3_bucket ||
+        inference_file === "s3" ||
+        inference_file === "s3_bucket" ||
+        inference_file === "S3 Bucket" ||
+        inference_file === "[Attached S3 Bucket]";
+
+    const isImagesFolder = targetFilePath && (
+        targetFilePath.endsWith("/images") || targetFilePath.endsWith("\\images")
+    );
+
+    let useS3 = isS3Selection;
+    let bucket = null;
+
+    try {
+        if (queries && queries.managed && typeof queries.managed.getBucket === "function") {
+            const bucketResult = await queries.managed.getBucket(PName, Admin);
+            bucket = bucketResult && bucketResult.row;
+        }
+    } catch (e) {
+        // Ignore error if getBucket is unmocked or fails
+    }
 
+    if (!useS3 && isImagesFolder && bucket) {
+        const localImages = fs.existsSync(targetFilePath) ? fs.readdirSync(targetFilePath) : [];
+        if (localImages.length === 0) {
+            useS3 = true;
+        }
+    }
+
+    // S3 Bucket stream mode (JIT fetch right before inference run execution)
+    if (useS3) {
         if (!bucket) {
             throw new Error("No S3 bucket attached to this project");
         }
diff --git a/views/training/inceptionSettings.ejs b/views/training/inceptionSettings.ejs
index 117dcee1..57b1fb64 100644
--- a/views/training/inceptionSettings.ejs
+++ b/views/training/inceptionSettings.ejs
@@ -166,7 +166,9 @@
                                         <h5><label for="Inference">Inference Files:</label></h5>
                                         <select name="Inference" id="Inference">
                                             <% for (var i = 0; i < global_inference_upload.length; i++) { %>
-                                                <option><%= global_inference_upload[i] %></option>
+                                                <option value="<%= global_inference_upload[i] %>">
+                                                    <%= global_inference_upload[i] === 's3' || global_inference_upload[i] === 's3_bucket' ? '[Attached S3 Bucket]' : global_inference_upload[i] %>
+                                                </option>
                                             <% } %>
                                         </select>
                                     </div>
diff --git a/views/training/megadetectorSettings.ejs b/views/training/megadetectorSettings.ejs
index 97ff30d4..3c21abbf 100644
--- a/views/training/megadetectorSettings.ejs
+++ b/views/training/megadetectorSettings.ejs
@@ -85,8 +85,8 @@
                                                 <h5><label for="Inference_md">Inference Files:</label></h5>
                                                 <select name="Inference_md" id="Inference_md" style="width:100%">
                                                     <% for(var i=0; i<global_inference_upload.length; i++) {%>
-                                                        <option>
-                                                            <%= global_inference_upload[i]%>
+                                                        <option value="<%= global_inference_upload[i] %>">
+                                                            <%= global_inference_upload[i] === 's3' || global_inference_upload[i] === 's3_bucket' ? '[Attached S3 Bucket]' : global_inference_upload[i] %>
                                                         </option>
                                                         <% } %>
                                                 </select>
diff --git a/views/training/yolovXInferenceSettings.ejs b/views/training/yolovXInferenceSettings.ejs
index 2fbfb599..9f54499d 100644
--- a/views/training/yolovXInferenceSettings.ejs
+++ b/views/training/yolovXInferenceSettings.ejs
@@ -155,10 +155,10 @@
                                                 <h5><label for="upload-command">Inference Files:</label></h5>
                                                 <select name="Inference" id="Inference" style="width:100%">
                                                     <% for(var i=0; i<global_inference_upload.length; i++) {%>
-                                                        <option>
-                                                            <%= global_inference_upload[i]%>
+                                                        <option value="<%= global_inference_upload[i] %>">
+                                                            <%= global_inference_upload[i] === 's3' || global_inference_upload[i] === 's3_bucket' ? '[Attached S3 Bucket]' : global_inference_upload[i] %>
                                                         </option>
-                                                        <% } %>
+                                                    <% } %>
                                                 </select>
                                             </div>
                                         </div>

From f37f40792da2b458ce62f8da76d0a776f0475cb2 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:54:32 -0700
Subject: [PATCH 14/18] fix(inference): add existence check before reading
 image file in yoloInference to handle streaming mode images

Co-authored-by: multica-agent <github@multica.ai>
---
 routes/inference/yoloInference.js |  9 ++++++---
 runs/summary.json                 | 12 ++++++------
 tests/run_summary.md              |  2 +-
 tests/summary.json                | 12 ++++++------
 4 files changed, 19 insertions(+), 16 deletions(-)

diff --git a/routes/inference/yoloInference.js b/routes/inference/yoloInference.js
index 813b5acb..f302805d 100644
--- a/routes/inference/yoloInference.js
+++ b/routes/inference/yoloInference.js
@@ -79,9 +79,12 @@ async function yoloInference(req, res) {
         var dictImagesLabels = {};
 
         for (var i = 0; i < existingImages.rows.length; i++) {
-            var img = fs.readFileSync(
-                `${imagesPath}/${existingImages.rows[i].IName}`,
-            ),
+            const imgPath = path.join(imagesPath, existingImages.rows[i].IName);
+            if (!fs.existsSync(imgPath)) {
+                continue;
+            }
+
+            var img = fs.readFileSync(imgPath),
                 imgData = probe.sync(img),
                 imgW = imgData.width,
                 imgH = imgData.height;
diff --git a/runs/summary.json b/runs/summary.json
index b2d2e921..94975623 100644
--- a/runs/summary.json
+++ b/runs/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": null,
-  "generatedAt": "2026-08-18T23:51:05.339Z",
+  "generatedAt": "2026-08-18T23:54:19.778Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -27,7 +27,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/runs/detect/train/detect_summary_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:51:05.338Z",
+      "generatedAt": "2026-08-18T23:54:19.765Z",
       "artifactCount": 1,
       "imageCount": 0,
       "config": {
@@ -47,10 +47,10 @@
       "visualPlots": [],
       "findings": [],
       "recommendations": [],
-      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:05.338Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
-      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:05.338Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
+      "markdownSummary": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:54:19.765Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n",
+      "summaryMd": "# Run Summary: detect_summary_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:54:19.765Z\n- **Artifact Count**: 1 files (0 images)\n\n## Executive Summary & Findings\n- No summary findings recorded.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"model\": \"yolo11n.pt\",\n  \"epochs\": 10\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:51:05.339Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
-  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:51:05.339Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
+  "markdownSummary": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:54:19.778Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n",
+  "summaryMd": "# Run Summary: all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: All Projects\n- **Generated At**: 2026-08-18T23:54:19.778Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s).\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 0.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 0\n- **Best Overall mAP@50**: 0.00%\n- **Top Performing Run**: `None`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `detect_summary_run` | training | 0 | N/A | N/A |\n\n## AI Recommendations & Next Steps\n- 💡 Consider increasing training epochs or fine-tuning hyperparameters for improved accuracy.\n"
 }
\ No newline at end of file
diff --git a/tests/run_summary.md b/tests/run_summary.md
index e5cdc69b..6bcba8e2 100644
--- a/tests/run_summary.md
+++ b/tests/run_summary.md
@@ -2,7 +2,7 @@
 
 - **Type**: AGGREGATED ALL-RUNS REPORT
 - **Project**: tmp_test_run
-- **Generated At**: 2026-08-18T23:51:04.138Z
+- **Generated At**: 2026-08-18T23:54:18.084Z
 - **Total Runs Analyzed**: 1 (1 training, 0 inference)
 
 ## Executive Summary & Findings
diff --git a/tests/summary.json b/tests/summary.json
index c3f57a4d..3d1fd6a1 100644
--- a/tests/summary.json
+++ b/tests/summary.json
@@ -4,7 +4,7 @@
   "runType": "aggregated",
   "isAggregated": true,
   "projectName": "tmp_test_run",
-  "generatedAt": "2026-08-18T23:51:04.138Z",
+  "generatedAt": "2026-08-18T23:54:18.084Z",
   "totalRuns": 1,
   "trainingRunCount": 1,
   "inferenceRunCount": 0,
@@ -28,7 +28,7 @@
       "runDir": "/home/fc/multica_workspaces/09feba42-72e2-4319-888e-970a8e60ff21/b772ab26/workdir/Njobvu-AI/tests/tmp_test_run",
       "runType": "training",
       "isAggregated": false,
-      "generatedAt": "2026-08-18T23:51:04.137Z",
+      "generatedAt": "2026-08-18T23:54:18.083Z",
       "artifactCount": 4,
       "imageCount": 1,
       "config": {
@@ -69,10 +69,10 @@
       "recommendations": [
         "Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment."
       ],
-      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:04.137Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
-      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:51:04.137Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
+      "markdownSummary": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:54:18.083Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n",
+      "summaryMd": "# Run Summary: tmp_test_run\n\n- **Run Type**: TRAINING\n- **Execution Status**: Unknown\n- **Generated At**: 2026-08-18T23:54:18.083Z\n- **Artifact Count**: 4 files (1 images)\n\n## Executive Summary & Findings\n- Completed 3 epochs of training.\n- Achieved peak mAP@50 of 85.00% (Epoch 3).\n\n## Performance & Metrics Analysis\n- **Loss Reduction**: N/A (Initial: null, Final: null)\n- **Loss Trajectory**: stable\n```json\n{\n  \"totalEpochs\": 3,\n  \"bestMap50\": 0.85,\n  \"bestMap50Epoch\": 3,\n  \"bestMap50_95\": 0,\n  \"epoch\": 3,\n  \"mAP50\": 0.85\n}\n```\n\n## AI Recommendations & Next Steps\n- 💡 Model achieved strong detection accuracy (mAP@50 > 80%). Ready for validation and deployment.\n\n## Configuration & Hyperparameters\n```json\n{\n  \"epochs\": 10,\n  \"batch\": 16,\n  \"model\": \"yolov8n.pt\"\n}\n```\n"
     }
   ],
-  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:51:04.138Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
-  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:51:04.138Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
+  "markdownSummary": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:54:18.084Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n",
+  "summaryMd": "# Run Summary: tmp_test_run_all_runs_summary\n\n- **Type**: AGGREGATED ALL-RUNS REPORT\n- **Project**: tmp_test_run\n- **Generated At**: 2026-08-18T23:54:18.084Z\n- **Total Runs Analyzed**: 1 (1 training, 0 inference)\n\n## Executive Summary & Findings\n- Aggregated analysis for 1 run(s) in project 'tmp_test_run'.\n- Training runs: 1, Inference runs: 0.\n- Total epochs trained across all runs: 3.\n- Top performing training run: 'tmp_test_run' with peak mAP@50 of 85.00%.\n\n## Aggregate Metrics\n- **Total Epochs Trained**: 3\n- **Best Overall mAP@50**: 85.00%\n- **Top Performing Run**: `tmp_test_run`\n\n## Individual Run Breakdown\n| Run Name | Type | Images | Best mAP@50 | Total Epochs |\n| --- | --- | --- | --- | --- |\n| `tmp_test_run` | training | 1 | 85.00% | 3 |\n\n## AI Recommendations & Next Steps\n- 💡 Run 'tmp_test_run' demonstrated high detection performance (mAP@50 > 80%). Recommended for deployment.\n"
 }
\ No newline at end of file

From 783a0f0a2ff14501fab71ad5275b57d1b026b706 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Tue, 18 Aug 2026 16:59:22 -0700
Subject: [PATCH 15/18] fix

---
 routes/pages/getYoloXInferenceSettingsPage.js | 38 ++++++++++---------
 routes/pages/getYoloXSettingsPage.js          | 24 ++++++------
 2 files changed, 32 insertions(+), 30 deletions(-)

diff --git a/routes/pages/getYoloXInferenceSettingsPage.js b/routes/pages/getYoloXInferenceSettingsPage.js
index 5e17c568..b6f8be15 100644
--- a/routes/pages/getYoloXInferenceSettingsPage.js
+++ b/routes/pages/getYoloXInferenceSettingsPage.js
@@ -1,3 +1,5 @@
+const queries = require("../../queries/queries");
+
 async function getYoloXInferencePage(req, res) {
     const readdir = util.promisify(fs.readdir);
     const readFile = util.promisify(fs.readFile);
@@ -48,12 +50,12 @@ async function getYoloXInferencePage(req, res) {
         fs.mkdirSync(log_path);
         fs.mkdirSync(python_path);
         fs.mkdirSync(weights_path);
-        fs.writeFile(python_path_file, "", function (err) {
+        fs.writeFile(python_path_file, "", function(err) {
             if (err) {
                 global.logger.error(err);
             }
         });
-        fs.writeFile(yolovx_path_file, "", function (err) {
+        fs.writeFile(yolovx_path_file, "", function(err) {
             if (err) {
                 global.logger.error(err);
             }
@@ -61,7 +63,7 @@ async function getYoloXInferencePage(req, res) {
     } else if (!fs.existsSync(weights_path)) {
         fs.mkdirSync(weights_path);
     } else if (!fs.existsSync(yolovx_path_file)) {
-        fs.writeFile(yolovx_path_file, "", function (err) {
+        fs.writeFile(yolovx_path_file, "", function(err) {
             if (err) {
                 global.logger.error(err);
             }
@@ -77,10 +79,10 @@ async function getYoloXInferencePage(req, res) {
     });
 
     // create async database object functions
-    tdb.getAsync = function (sql) {
+    tdb.getAsync = function(sql) {
         var that = this;
-        return new Promise(function (resolve, reject) {
-            that.get(sql, function (err, row) {
+        return new Promise(function(resolve, reject) {
+            that.get(sql, function(err, row) {
                 if (err) {
                     global.logger.error("runAsync ERROR!", err)
                     reject(err);
@@ -90,10 +92,10 @@ async function getYoloXInferencePage(req, res) {
             global.logger.error(err);
         });
     };
-    tdb.allAsync = function (sql) {
+    tdb.allAsync = function(sql) {
         var that = this;
-        return new Promise(function (resolve, reject) {
-            that.all(sql, function (err, row) {
+        return new Promise(function(resolve, reject) {
+            that.all(sql, function(err, row) {
                 if (err) {
                     global.logger.error("runAsync ERROR!", err)
                     reject(err);
@@ -106,19 +108,19 @@ async function getYoloXInferencePage(req, res) {
 
     var results1 = await db.getAsync(
         "SELECT * FROM `Projects` WHERE PName = '" +
-            PName +
-            "' AND Admin = '" +
-            admin +
-            "'",
+        PName +
+        "' AND Admin = '" +
+        admin +
+        "'",
     );
     var results2 = await tdb.allAsync("SELECT * FROM `Classes`");
 
     var acc = await db.allAsync(
         "SELECT * FROM `Access` WHERE PName = '" +
-            PName +
-            "' AND Admin = '" +
-            admin +
-            "'",
+        PName +
+        "' AND Admin = '" +
+        admin +
+        "'",
     );
     var access = [];
     for (var i = 0; i < acc.length; i++) {
@@ -223,7 +225,7 @@ async function getYoloXInferencePage(req, res) {
     }
 
     // close the database
-    tdb.close(function (err) {
+    tdb.close(function(err) {
         if (err) {
             global.logger.error(err);
         } else {
diff --git a/routes/pages/getYoloXSettingsPage.js b/routes/pages/getYoloXSettingsPage.js
index 5700481a..c113b5f6 100644
--- a/routes/pages/getYoloXSettingsPage.js
+++ b/routes/pages/getYoloXSettingsPage.js
@@ -50,12 +50,12 @@ async function getYoloXSettingsPage(req, res) {
         fs.mkdirSync(log_path);
         fs.mkdirSync(python_path);
         fs.mkdirSync(weights_path);
-        fs.writeFile(python_path_file, "", function (err) {
+        fs.writeFile(python_path_file, "", function(err) {
             if (err) {
                 global.logger.error(err);
             }
         });
-        fs.writeFile(yolovx_path_file, "", function (err) {
+        fs.writeFile(yolovx_path_file, "", function(err) {
             if (err) {
                 global.logger.error(err);
             }
@@ -63,7 +63,7 @@ async function getYoloXSettingsPage(req, res) {
     } else if (!fs.existsSync(weights_path)) {
         fs.mkdirSync(weights_path);
     } else if (!fs.existsSync(yolovx_path_file)) {
-        fs.writeFile(yolovx_path_file, "", function (err) {
+        fs.writeFile(yolovx_path_file, "", function(err) {
             if (err) {
                 global.logger.error(err);
             }
@@ -79,10 +79,10 @@ async function getYoloXSettingsPage(req, res) {
     });
 
     // create async database object functions
-    tdb.getAsync = function (sql) {
+    tdb.getAsync = function(sql) {
         var that = this;
-        return new Promise(function (resolve, reject) {
-            that.get(sql, function (err, row) {
+        return new Promise(function(resolve, reject) {
+            that.get(sql, function(err, row) {
                 if (err) {
                     global.logger.error("runAsync ERROR!", err)
                     reject(err);
@@ -92,10 +92,10 @@ async function getYoloXSettingsPage(req, res) {
             global.logger.error(err);
         });
     };
-    tdb.allAsync = function (sql) {
+    tdb.allAsync = function(sql) {
         var that = this;
-        return new Promise(function (resolve, reject) {
-            that.all(sql, function (err, row) {
+        return new Promise(function(resolve, reject) {
+            that.all(sql, function(err, row) {
                 if (err) {
                     global.logger.error("runAsync ERROR!", err)
                     reject(err);
@@ -121,7 +121,7 @@ async function getYoloXSettingsPage(req, res) {
         var countsResult = await queries.project.getClassLabelCounts(project_path);
 
         if (countsResult && countsResult.rows) {
-            countsResult.rows.forEach(function (row) {
+            countsResult.rows.forEach(function(row) {
                 classLabelCounts[row.CName] = row.labelCount;
             });
         }
@@ -129,7 +129,7 @@ async function getYoloXSettingsPage(req, res) {
         global.logger.error(err);
     }
 
-    results2 = results2.map(function (cls) {
+    results2 = results2.map(function(cls) {
         return Object.assign({}, cls, {
             labelCount: classLabelCounts[cls.CName] || 0,
         });
@@ -239,7 +239,7 @@ async function getYoloXSettingsPage(req, res) {
     }
 
     // close the database
-    tdb.close(function (err) {
+    tdb.close(function(err) {
         if (err) {
             global.logger.error(err);
         } else {

From 3cc8aaef2fd24f0c0d918ea6558c507844b12d49 Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Wed, 19 Aug 2026 11:05:15 -0700
Subject: [PATCH 16/18] feat(s3): add JIT image pulling and cleanup for
 streamed S3 training runs

Co-authored-by: multica-agent <github@multica.ai>
---
 package-lock.json                             |  26 ++-
 routes/training/run.js                        |  22 ++-
 routes/training/yoloRun.js                    |  15 +-
 .../jitTrainingImagesPipeline.test.js         | 175 ++++++++++++++++++
 tests/unit/jitTrainingImages.test.js          | 127 +++++++++++++
 utils/jitTrainingImages.js                    | 111 +++++++++++
 6 files changed, 470 insertions(+), 6 deletions(-)
 create mode 100644 tests/integration/jitTrainingImagesPipeline.test.js
 create mode 100644 tests/unit/jitTrainingImages.test.js
 create mode 100644 utils/jitTrainingImages.js

diff --git a/package-lock.json b/package-lock.json
index 3ddbfdbe..81e2e1bd 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -407,7 +407,6 @@
       "integrity": "sha512-bXYxrXFubeYdvB0NhD/NBB3Qi6aZeV20GOWVI47t2dkecCEoneR4NPVcb7abpXDEvejgrUfFtG6vG/zxAKmg+g==",
       "dev": true,
       "license": "MIT",
-      "peer": true,
       "dependencies": {
         "@ampproject/remapping": "^2.2.0",
         "@babel/code-frame": "^7.27.1",
@@ -2382,7 +2381,6 @@
       "version": "19.0.6",
       "resolved": "https://registry.npmjs.org/@types/react/-/react-19.0.6.tgz",
       "integrity": "sha512-gIlMztcTeDgXCUj0vCBOqEuSEhX//63fW9SZtCJ+agxoQTOklwDfiEMlTWn4mR/C/UK5VHlpwsCsOyf7/hc4lw==",
-      "peer": true,
       "dependencies": {
         "csstype": "^3.0.2"
       }
@@ -3205,7 +3203,6 @@
         }
       ],
       "license": "MIT",
-      "peer": true,
       "dependencies": {
         "caniuse-lite": "^1.0.30001718",
         "electron-to-chromium": "^1.5.160",
@@ -4392,6 +4389,29 @@
         "node": ">= 0.8"
       }
     },
+    "node_modules/encoding": {
+      "version": "0.1.13",
+      "resolved": "https://registry.npmjs.org/encoding/-/encoding-0.1.13.tgz",
+      "integrity": "sha512-ETBauow1T35Y/WZMkio9jiM0Z5xjHHmJ4XmjZOq1l/dXz3lr2sRn87nJy20RupqSh1F2m3HHPSp8ShIPQJrJ3A==",
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "iconv-lite": "^0.6.2"
+      }
+    },
+    "node_modules/encoding/node_modules/iconv-lite": {
+      "version": "0.6.3",
+      "resolved": "https://registry.npmjs.org/iconv-lite/-/iconv-lite-0.6.3.tgz",
+      "integrity": "sha512-4fCk79wshMdzMp2rH06qWrJE4iolqLhCUH+OiuIgU++RB0+94NlDL81atO7GX55uUKueo0txHNtvEyI6D7WdMw==",
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "safer-buffer": ">= 2.1.2 < 3.0.0"
+      },
+      "engines": {
+        "node": ">=0.10.0"
+      }
+    },
     "node_modules/end-of-stream": {
       "version": "1.4.4",
       "resolved": "https://registry.npmjs.org/end-of-stream/-/end-of-stream-1.4.4.tgz",
diff --git a/routes/training/run.js b/routes/training/run.js
index 3c2793c9..0041ca36 100644
--- a/routes/training/run.js
+++ b/routes/training/run.js
@@ -1,5 +1,11 @@
+const path = require("path");
+const fs = require("fs");
 const queries = require("../../queries/queries");
 const formatRunOptionsHeader = require("../../utils/formatRunOptionsHeader");
+const {
+    ensureTrainingImagesLocal,
+    cleanupJitTrainingImages,
+} = require("../../utils/jitTrainingImages");
 
 async function run(req, res) {
     const { exec } = require("child_process");
@@ -27,7 +33,7 @@ async function run(req, res) {
         options = "EMPTY";
     }
     global.logger.debug("options: ", options);
-    var publicPath = currentPath,
+    var publicPath = global.currentPath || (typeof currentPath !== "undefined" ? currentPath : (process.cwd() + "/")),
         mainPath = publicPath + "public/projects/", // $LABELING_TOOL_PATH/public/projects/
         projectPath = mainPath + Admin + "-" + PName, // $LABELING_TOOL_PATH/public/projects/project_name
         imagesPath = projectPath + "/images", // $LABELING_TOOL_PATH/public/projects/project_name/images
@@ -59,11 +65,22 @@ async function run(req, res) {
         if (err) throw err;
     });
 
+    let existingImages;
+    let jitDownloadedFiles = [];
+    try {
+        existingImages = await queries.project.getAllImages(projectPath);
+        jitDownloadedFiles = await ensureTrainingImagesLocal(PName, Admin, projectPath, existingImages.rows);
+    } catch (err) {
+        global.logger.error("Error fetching streamed S3 images JIT for run:", err);
+        return res.status(500).send("Error fetching streamed S3 images for training: " + err.message);
+    }
+
     let existingLabels;
     try {
         existingLabels = await queries.project.getAllLabels(projectPath);
     } catch (err) {
         global.logger.error(err);
+        await cleanupJitTrainingImages(jitDownloadedFiles);
         return res.status(500).send("Error fetching labels");
     }
     const labels = existingLabels.rows;
@@ -168,8 +185,9 @@ async function run(req, res) {
         global.logger.debug("stdout: ", stdout);
         global.logger.debug("stderr: ", stderr);
         global.logger.debug("err: ", err);
-        fs.writeFile(`${runPath}/done.log`, success, (err) => {
+        fs.writeFile(`${runPath}/done.log`, success, async (err) => {
             if (err) throw err;
+            await cleanupJitTrainingImages(jitDownloadedFiles);
         });
     });
 
diff --git a/routes/training/yoloRun.js b/routes/training/yoloRun.js
index cdeb7aac..a595756e 100644
--- a/routes/training/yoloRun.js
+++ b/routes/training/yoloRun.js
@@ -6,6 +6,10 @@ const probe = require("probe-image-size");
 const os = require("os");
 const sharp = require("sharp");
 const formatRunOptionsHeader = require("../../utils/formatRunOptionsHeader");
+const {
+    ensureTrainingImagesLocal,
+    cleanupJitTrainingImages,
+} = require("../../utils/jitTrainingImages");
 
 // Function to detect the best available device for YOLO training
 async function detectBestDevice() {
@@ -486,6 +490,14 @@ async function yoloRun(req, res) {
         targetImages = targetImages.slice(0, maxImages);
     }
 
+    let jitDownloadedFiles = [];
+    try {
+        jitDownloadedFiles = await ensureTrainingImagesLocal(PName, Admin, projectPath, targetImages);
+    } catch (err) {
+        global.logger.error("Error ensuring local images for training JIT:", err);
+        return res.status(500).send("Error fetching streamed S3 images for training: " + err.message);
+    }
+
     // Parse Train : Validate : Test split ratio
     let trainPct = parseFloat(req.body.TrainingPercent || req.body.train_percent || 70);
     let valPct = parseFloat(req.body.ValPercent || req.body.val_percent);
@@ -903,7 +915,7 @@ async function yoloRun(req, res) {
     fs.writeFileSync(`${absDarknetProjectRun}/${log}`, `${runOptionsHeader}${cmd}`);
 
     const bufferSizeMult = (global.configFile && global.configFile.training_max_buffer_size) || (typeof configFile !== "undefined" && configFile.training_max_buffer_size) || 1;
-    exec(cmd, { maxBuffer: 1024 * 1024 * 1024 * bufferSizeMult }, (err, stdout, stderr) => {
+    exec(cmd, { maxBuffer: 1024 * 1024 * 1024 * bufferSizeMult }, async (err, stdout, stderr) => {
         if (stdout) {
             global.logger.debug("STDOUT:", stdout);
             fs.appendFile(`${absDarknetProjectRun}/${log}`, stdout, (err) => {
@@ -940,6 +952,7 @@ async function yoloRun(req, res) {
         }
 
         fs.writeFileSync(`${runPath}/done.log`, success);
+        await cleanupJitTrainingImages(jitDownloadedFiles);
     });
 
 
diff --git a/tests/integration/jitTrainingImagesPipeline.test.js b/tests/integration/jitTrainingImagesPipeline.test.js
new file mode 100644
index 00000000..23fd2b57
--- /dev/null
+++ b/tests/integration/jitTrainingImagesPipeline.test.js
@@ -0,0 +1,175 @@
+jest.mock('decompress-zip', () => jest.fn());
+jest.mock('decompress-zip/lib/extractors', () => ({
+  folder: jest.fn(),
+}));
+jest.mock('ffmpeg', () => jest.fn());
+jest.mock('sharp', () => jest.fn());
+jest.mock('unzipper', () => jest.fn());
+jest.mock('probe-image-size', () => ({
+  sync: jest.fn().mockReturnValue({ width: 640, height: 480 }),
+}));
+
+jest.mock('child_process', () => ({
+  exec: jest.fn((cmd, opts, cb) => {
+    const callback = typeof opts === 'function' ? opts : cb;
+    if (typeof callback === 'function') {
+      process.nextTick(() => callback(null, 'OK', ''));
+    }
+  }),
+}));
+
+const request = require('supertest');
+const fs = require('fs');
+const path = require('path');
+const queries = require('../../queries/queries');
+const s3Client = require('../../utils/s3Client');
+
+jest.mock('../../queries/queries', () => ({
+  managed: {
+    getBucket: jest.fn(),
+  },
+  project: {
+    getAllImages: jest.fn(),
+    getAllClasses: jest.fn(),
+    getLabelsForImageName: jest.fn(),
+    getAllLabels: jest.fn(),
+  },
+}));
+
+jest.mock('../../utils/s3Client', () => ({
+  buildS3Client: jest.fn(() => ({ fakeS3Client: true })),
+  downloadObjectToFile: jest.fn(),
+}));
+
+const app = require('../../app');
+
+describe('JIT Training Images Pipeline Integration Test', () => {
+  beforeEach(() => {
+    jest.clearAllMocks();
+    jest.spyOn(process, 'chdir').mockImplementation(() => {});
+
+    queries.project.getAllImages.mockResolvedValue({
+      rows: [
+        { IName: 'local_img.jpg', Source: null, SourceKey: null },
+        { IName: 'stream_img.jpg', Source: 's3', SourceKey: 's3/stream_img.jpg' },
+      ],
+    });
+
+    queries.project.getAllClasses.mockResolvedValue({
+      rows: [{ CName: 'car' }],
+    });
+
+    queries.project.getLabelsForImageName.mockResolvedValue({
+      rows: [{ LID: 1, CName: 'car', X: 10, Y: 20, W: 30, H: 40 }],
+    });
+
+    queries.project.getAllLabels.mockResolvedValue({
+      rows: [{ IName: 'stream_img.jpg', CName: 'car', X: 10, Y: 20, W: 30, H: 40 }],
+    });
+
+    queries.managed.getBucket.mockResolvedValue({
+      row: {
+        BucketName: 'test-bucket',
+        Region: 'us-east-1',
+        AccessKeyId: 'key',
+        SecretAccessKey: 'secret',
+      },
+    });
+
+    s3Client.downloadObjectToFile.mockResolvedValue();
+  });
+
+  test('POST /yolo-run triggers JIT download for missing streamed S3 images and unlinks them post-run', async () => {
+    const unlinkSpy = jest.spyOn(fs.promises, 'unlink').mockResolvedValue();
+
+    jest.spyOn(fs, 'existsSync').mockImplementation((p) => {
+      if (typeof p === 'string' && p.includes('stream_img.jpg')) {
+        return false;
+      }
+      return true;
+    });
+    jest.spyOn(fs, 'mkdirSync').mockImplementation(() => {});
+    jest.spyOn(fs, 'copyFileSync').mockImplementation(() => {});
+    jest.spyOn(fs, 'copyFile').mockImplementation((src, dest, cb) => { if (cb) cb(null); });
+    jest.spyOn(fs, 'readFileSync').mockReturnValue(Buffer.from('fake image data'));
+    jest.spyOn(fs, 'writeFileSync').mockImplementation(() => {});
+    jest.spyOn(fs, 'appendFile').mockImplementation((p, data, cb) => { if (cb) cb(null); });
+    jest.spyOn(fs, 'writeFile').mockImplementation((p, data, cb) => { if (cb) cb(null); });
+    jest.spyOn(fs.promises, 'symlink').mockResolvedValue();
+
+    const response = await request(app)
+      .post('/yolo-run')
+      .set('Cookie', ['Username=testuser'])
+      .send({
+        PName: 'testproj',
+        Admin: 'testuser',
+        yolo_task: 'detect',
+        selected_classes: JSON.stringify(['car']),
+        TrainingPercent: 70,
+        weights: 'best.pt',
+        yolovx_path: '/usr/local/bin/yolo',
+      });
+
+    expect(response.status).toBe(200);
+    expect(response.body).toEqual({ Success: 'YOLO Training Started' });
+
+    // Verify JIT fetch downloaded the missing streamed image
+    expect(s3Client.downloadObjectToFile).toHaveBeenCalledWith(
+      { fakeS3Client: true },
+      'test-bucket',
+      's3/stream_img.jpg',
+      expect.stringContaining('stream_img.jpg'),
+    );
+
+    // Wait for async background cleanup in exec callback
+    await new Promise((r) => setTimeout(r, 50));
+
+    // Verify cleanup unlinked the JIT-downloaded image when training completed
+    expect(unlinkSpy).toHaveBeenCalledWith(expect.stringContaining('stream_img.jpg'));
+    // Ensure local_img.jpg was not unlinked
+    expect(unlinkSpy).not.toHaveBeenCalledWith(expect.stringContaining('local_img.jpg'));
+  });
+
+  test('POST /run triggers JIT download for missing streamed S3 images and unlinks them post-run', async () => {
+    const unlinkSpy = jest.spyOn(fs.promises, 'unlink').mockResolvedValue();
+
+    jest.spyOn(fs, 'existsSync').mockImplementation((p) => {
+      if (typeof p === 'string' && p.includes('stream_img.jpg')) {
+        return false;
+      }
+      return true;
+    });
+    jest.spyOn(fs, 'mkdirSync').mockImplementation(() => {});
+    jest.spyOn(fs, 'copyFileSync').mockImplementation(() => {});
+    jest.spyOn(fs, 'writeFile').mockImplementation((p, data, cb) => { if (cb) cb(null); });
+
+    const response = await request(app)
+      .post('/run')
+      .set('Cookie', ['Username=testuser'])
+      .send({
+        PName: 'testproj',
+        Admin: 'testuser',
+        script: 'train.py',
+        python_path: '/usr/bin/python3',
+        TrainingPercent: 70,
+        weights: 'weights.h5',
+      });
+
+    expect(response.status).toBe(200);
+    expect(response.body).toEqual({ Success: 'Training Started' });
+
+    // Verify JIT fetch downloaded missing streamed S3 image
+    expect(s3Client.downloadObjectToFile).toHaveBeenCalledWith(
+      { fakeS3Client: true },
+      'test-bucket',
+      's3/stream_img.jpg',
+      expect.stringContaining('stream_img.jpg'),
+    );
+
+    // Wait for async background cleanup in exec callback
+    await new Promise((r) => setTimeout(r, 50));
+
+    // Verify cleanup unlinked JIT downloaded image
+    expect(unlinkSpy).toHaveBeenCalledWith(expect.stringContaining('stream_img.jpg'));
+  });
+});
diff --git a/tests/unit/jitTrainingImages.test.js b/tests/unit/jitTrainingImages.test.js
new file mode 100644
index 00000000..a6dfabd3
--- /dev/null
+++ b/tests/unit/jitTrainingImages.test.js
@@ -0,0 +1,127 @@
+const fs = require('fs');
+const path = require('path');
+const queries = require('../../queries/queries');
+const s3Client = require('../../utils/s3Client');
+const {
+    ensureTrainingImagesLocal,
+    cleanupJitTrainingImages,
+} = require('../../utils/jitTrainingImages');
+
+jest.mock('../../queries/queries', () => ({
+    managed: {
+        getBucket: jest.fn(),
+    },
+    project: {
+        getAllImages: jest.fn(),
+    },
+}));
+
+jest.mock('../../utils/s3Client', () => ({
+    buildS3Client: jest.fn(() => ({ fakeS3Client: true })),
+    downloadObjectToFile: jest.fn(),
+}));
+
+describe('utils/jitTrainingImages', () => {
+    beforeEach(() => {
+        jest.clearAllMocks();
+        global.logger = { debug: jest.fn(), error: jest.fn(), info: jest.fn() };
+    });
+
+    describe('ensureTrainingImagesLocal', () => {
+        it('returns empty array when targetImages is empty or undefined', async () => {
+            const downloaded = await ensureTrainingImagesLocal('proj', 'user', '/path/to/proj', []);
+            expect(downloaded).toEqual([]);
+            expect(s3Client.downloadObjectToFile).not.toHaveBeenCalled();
+        });
+
+        it('skips images that already exist locally on disk', async () => {
+            const projectPath = '/projects/user-proj';
+            const targetImages = [
+                { IName: 'local1.jpg', Source: 's3', SourceKey: 'keys/local1.jpg' },
+            ];
+
+            jest.spyOn(fs, 'existsSync').mockReturnValue(true);
+
+            const downloaded = await ensureTrainingImagesLocal('proj', 'user', projectPath, targetImages);
+
+            expect(downloaded).toEqual([]);
+            expect(s3Client.downloadObjectToFile).not.toHaveBeenCalled();
+        });
+
+        it('downloads missing S3-backed images JIT and returns their file paths', async () => {
+            const projectPath = '/projects/user-proj';
+            const targetImages = [
+                { IName: 'stream1.jpg', Source: 's3', SourceKey: 'keys/stream1.jpg' },
+                { IName: 'local1.jpg', Source: null, SourceKey: null },
+            ];
+
+            jest.spyOn(fs, 'existsSync').mockImplementation((p) => {
+                if (p.includes('stream1.jpg')) return false;
+                return true;
+            });
+            jest.spyOn(fs, 'mkdirSync').mockImplementation(() => {});
+
+            queries.managed.getBucket.mockResolvedValueOnce({
+                row: {
+                    BucketName: 'my-bucket',
+                    Region: 'us-east-1',
+                    AccessKeyId: 'AKIA...',
+                    SecretAccessKey: 'secret',
+                    Endpoint: '',
+                },
+            });
+            s3Client.downloadObjectToFile.mockResolvedValueOnce();
+
+            const downloaded = await ensureTrainingImagesLocal('proj', 'user', projectPath, targetImages);
+
+            const expectedPath = path.join(projectPath, 'images', 'stream1.jpg');
+            expect(downloaded).toEqual([expectedPath]);
+            expect(s3Client.buildS3Client).toHaveBeenCalledWith({
+                region: 'us-east-1',
+                accessKeyId: 'AKIA...',
+                secretAccessKey: 'secret',
+                endpoint: '',
+            });
+            expect(s3Client.downloadObjectToFile).toHaveBeenCalledWith(
+                { fakeS3Client: true },
+                'my-bucket',
+                'keys/stream1.jpg',
+                expectedPath,
+            );
+        });
+
+        it('throws an error if missing S3 images exist but no S3 bucket is attached', async () => {
+            const projectPath = '/projects/user-proj';
+            const targetImages = [
+                { IName: 'stream1.jpg', Source: 's3', SourceKey: 'keys/stream1.jpg' },
+            ];
+
+            jest.spyOn(fs, 'existsSync').mockReturnValue(false);
+            queries.managed.getBucket.mockResolvedValueOnce({ row: null });
+
+            await expect(
+                ensureTrainingImagesLocal('proj', 'user', projectPath, targetImages),
+            ).rejects.toThrow('S3 bucket configuration missing for project proj');
+        });
+    });
+
+    describe('cleanupJitTrainingImages', () => {
+        it('unlinks files present in the downloadedFiles array', async () => {
+            const downloadedFiles = ['/path/to/stream1.jpg', '/path/to/stream2.jpg'];
+
+            jest.spyOn(fs, 'existsSync').mockReturnValue(true);
+            jest.spyOn(fs.promises, 'unlink').mockResolvedValue();
+
+            await cleanupJitTrainingImages(downloadedFiles);
+
+            expect(fs.promises.unlink).toHaveBeenCalledWith('/path/to/stream1.jpg');
+            expect(fs.promises.unlink).toHaveBeenCalledWith('/path/to/stream2.jpg');
+        });
+
+        it('does nothing when downloadedFiles is empty', async () => {
+            jest.spyOn(fs.promises, 'unlink').mockResolvedValue();
+            await cleanupJitTrainingImages([]);
+            expect(fs.promises.unlink).not.toHaveBeenCalled();
+        });
+    });
+});
diff --git a/utils/jitTrainingImages.js b/utils/jitTrainingImages.js
new file mode 100644
index 00000000..7400a059
--- /dev/null
+++ b/utils/jitTrainingImages.js
@@ -0,0 +1,111 @@
+const fs = require("fs");
+const path = require("path");
+const queries = require("../queries/queries");
+const { buildS3Client, downloadObjectToFile } = require("./s3Client");
+
+/**
+ * Ensures all requested target images for a training run exist locally on disk.
+ * If any image is S3-backed (Source === 's3' && SourceKey) and missing from disk,
+ * it fetches the project S3 bucket credentials and downloads the image JIT.
+ *
+ * @param {string} projectName - Project name (PName)
+ * @param {string} admin - Admin / owner username
+ * @param {string} projectPath - Path to project directory
+ * @param {Array<Object|string>} targetImages - Array of image DB rows or image names
+ * @returns {Promise<Array<string>>} List of absolute file paths downloaded JIT by this call
+ */
+async function ensureTrainingImagesLocal(projectName, admin, projectPath, targetImages) {
+    if (!Array.isArray(targetImages) || targetImages.length === 0) {
+        return [];
+    }
+
+    const imagesPath = path.join(projectPath, "images");
+
+    // Retrieve DB image records if targetImages consists of plain strings or objects missing Source
+    let imageRows = targetImages;
+    const firstImg = targetImages[0];
+    if (typeof firstImg === "string" || !firstImg || typeof firstImg !== "object" || !("Source" in firstImg)) {
+        try {
+            const dbImagesResult = await queries.project.getAllImages(projectPath);
+            const dbImagesMap = new Map();
+            for (const row of (dbImagesResult && dbImagesResult.rows) || []) {
+                dbImagesMap.set(row.IName, row);
+            }
+
+            imageRows = targetImages.map((img) => {
+                const name = typeof img === "string" ? img : (img && img.IName);
+                return dbImagesMap.get(name) || (typeof img === "object" && img ? img : { IName: name });
+            });
+        } catch (err) {
+            if (global.logger && global.logger.error) {
+                global.logger.error("Error fetching image DB records for JIT check:", err);
+            }
+        }
+    }
+
+    const missingS3Images = [];
+    for (const img of imageRows) {
+        if (!img || !img.IName) continue;
+        const localPath = path.join(imagesPath, img.IName);
+        if (!fs.existsSync(localPath) && img.Source === "s3" && img.SourceKey) {
+            missingS3Images.push({ key: img.SourceKey, destPath: localPath });
+        }
+    }
+
+    if (missingS3Images.length === 0) {
+        return [];
+    }
+
+    const bucketResult = await queries.managed.getBucket(projectName, admin);
+    const bucket = bucketResult && bucketResult.row;
+    if (!bucket) {
+        throw new Error(`S3 bucket configuration missing for project ${projectName}`);
+    }
+
+    const s3Client = buildS3Client({
+        region: bucket.Region,
+        accessKeyId: bucket.AccessKeyId,
+        secretAccessKey: bucket.SecretAccessKey,
+        endpoint: bucket.Endpoint,
+    });
+
+    const downloadedFiles = [];
+    for (const { key, destPath } of missingS3Images) {
+        const dir = path.dirname(destPath);
+        if (!fs.existsSync(dir)) {
+            fs.mkdirSync(dir, { recursive: true });
+        }
+
+        await downloadObjectToFile(s3Client, bucket.BucketName, key, destPath);
+        downloadedFiles.push(destPath);
+    }
+
+    return downloadedFiles;
+}
+
+/**
+ * Unlinks files downloaded JIT for a training run.
+ * Leaves pre-existing local image files intact.
+ *
+ * @param {Array<string>} downloadedFiles - List of absolute file paths downloaded JIT
+ */
+async function cleanupJitTrainingImages(downloadedFiles) {
+    if (!Array.isArray(downloadedFiles) || downloadedFiles.length === 0) {
+        return;
+    }
+
+    for (const filePath of downloadedFiles) {
+        try {
+            await fs.promises.unlink(filePath);
+        } catch (err) {
+            if (err && err.code !== "ENOENT" && global.logger && global.logger.error) {
+                global.logger.error(`Error unlinking JIT training image ${filePath}:`, err);
+            }
+        }
+    }
+}
+
+module.exports = {
+    ensureTrainingImagesLocal,
+    cleanupJitTrainingImages,
+};

From 5a91ec76de92b28f47909ab1aee4279c79284dfa Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Thu, 20 Aug 2026 19:58:15 +0000
Subject: [PATCH 17/18] fix: resolve CI Run Tests failures in S3 and migration
 coverage

Co-authored-by: FerretCode <49322226+FerretCode@users.noreply.github.com>
---
 queries/projects/projects.js                  |   7 +-
 tests/integration/imageSourceColumn.test.js   |   5 +-
 tests/unit/getAnnotatePageS3Streaming.test.js | 109 ++++++++++--------
 3 files changed, 71 insertions(+), 50 deletions(-)

diff --git a/queries/projects/projects.js b/queries/projects/projects.js
index 3591da47..01aaf1df 100644
--- a/queries/projects/projects.js
+++ b/queries/projects/projects.js
@@ -231,9 +231,12 @@ module.exports = {
             // ADD COLUMN IF NOT EXISTS. SourceKey holds the literal S3 object key (which
             // may differ from IName once collisions are disambiguated), decoupled from the
             // display name.
-            const imageColumns = await db.all("PRAGMA table_info(Images)");
+            const imageColumnsResult = await db.all("PRAGMA table_info(Images)");
+            const imageColumns = Array.isArray(imageColumnsResult)
+                ? imageColumnsResult
+                : (imageColumnsResult && imageColumnsResult.rows) || [];
             const existingColumnNames = new Set(
-                (imageColumns.rows || []).map((column) => column.name),
+                imageColumns.map((column) => column.name),
             );
 
             const backfillColumns = [
diff --git a/tests/integration/imageSourceColumn.test.js b/tests/integration/imageSourceColumn.test.js
index cd0aebc6..d6a35b3c 100644
--- a/tests/integration/imageSourceColumn.test.js
+++ b/tests/integration/imageSourceColumn.test.js
@@ -79,7 +79,10 @@ describe('queries/projects migrateProjectDb Source/SourceKey column backfill', (
         await projects.project.migrateProjectDb('/projects/testuser-test-project');
 
         expect(mockRun).not.toHaveBeenCalledWith(
-            expect.stringContaining('ALTER TABLE Images ADD COLUMN'),
+            'ALTER TABLE Images ADD COLUMN Source VARCHAR DEFAULT NULL',
+        );
+        expect(mockRun).not.toHaveBeenCalledWith(
+            'ALTER TABLE Images ADD COLUMN SourceKey VARCHAR DEFAULT NULL',
         );
     });
 });
diff --git a/tests/unit/getAnnotatePageS3Streaming.test.js b/tests/unit/getAnnotatePageS3Streaming.test.js
index a1f3931f..17ffa093 100644
--- a/tests/unit/getAnnotatePageS3Streaming.test.js
+++ b/tests/unit/getAnnotatePageS3Streaming.test.js
@@ -15,8 +15,27 @@ jest.mock('../../utils/s3Client', () => ({
     buildS3Client: jest.fn(() => ({ fakeClient: true })),
     getObjectStream: jest.fn(),
 }));
+jest.mock('fs', () => ({
+    existsSync: jest.fn(),
+    readFileSync: jest.fn(),
+}));
+jest.mock('sqlite3', () => {
+    const Database = jest.fn();
+    return {
+        Database,
+        verbose: () => ({ Database }),
+    };
+});
+jest.mock('probe-image-size', () => {
+    const probe = jest.fn();
+    probe.sync = jest.fn();
+    return probe;
+});
 
 const { Readable } = require('stream');
+const fs = require('fs');
+const sqlite3 = require('sqlite3');
+const probe = require('probe-image-size');
 const queries = require('../../queries/queries');
 const s3Client = require('../../utils/s3Client');
 const getAnnotatePage = require('../../routes/pages/getAnnotatePage');
@@ -26,15 +45,17 @@ const getAnnotatePage = require('../../routes/pages/getAnnotatePage');
 // right after construction - only the raw callback methods are ever called.
 function makeFakeProjectDb({ classesRows = [], labelsRows = [], imagesRows = [], displayRow }) {
     return {
-        get: jest.fn((sql, cb) => {
-            if (sql.includes('display_id')) return cb(null, displayRow);
-            return cb(null, undefined);
+        get: jest.fn((sql, params, cb) => {
+            const callback = typeof params === 'function' ? params : cb;
+            if (sql.includes('display_id')) return callback(null, displayRow);
+            return callback(null, undefined);
         }),
-        all: jest.fn((sql, cb) => {
-            if (sql.includes('Classes')) return cb(null, classesRows);
-            if (sql.includes('Labels')) return cb(null, labelsRows);
-            if (sql.includes('Images')) return cb(null, imagesRows);
-            return cb(null, []);
+        all: jest.fn((sql, params, cb) => {
+            const callback = typeof params === 'function' ? params : cb;
+            if (sql.includes('Classes')) return callback(null, classesRows);
+            if (sql.includes('Labels')) return callback(null, labelsRows);
+            if (sql.includes('Images')) return callback(null, imagesRows);
+            return callback(null, []);
         }),
         each: jest.fn((sql, cb) => cb(null, undefined)),
         close: jest.fn((cb) => cb && cb(null)),
@@ -66,18 +87,17 @@ describe('getAnnotatePage - S3-backed image serving', () => {
     it('serves a "stream"-mode image with no local file via the on-demand S3 proxy, without touching disk', async () => {
         const imageRow = { IName: 'image1.jpg', reviewImage: 0, Source: 's3', SourceKey: 'images/image1.jpg' };
 
-        global.fs = { existsSync: jest.fn().mockReturnValue(false), readFileSync: jest.fn() };
-        global.sqlite3 = {
-            Database: jest.fn((dbPath, cb) => {
-                cb && cb(null);
-                return makeFakeProjectDb({
-                    imagesRows: [imageRow],
-                    displayRow: { IName: 'image1.jpg', display_id: 1 },
-                });
-            }),
-        };
-        global.probe = jest.fn().mockResolvedValue({ width: 400, height: 300 });
-        global.probe.sync = jest.fn();
+        fs.existsSync.mockReturnValue(false);
+        fs.readFileSync.mockReset();
+        sqlite3.Database.mockImplementation((dbPath, cb) => {
+            cb && cb(null);
+            return makeFakeProjectDb({
+                imagesRows: [imageRow],
+                displayRow: { IName: 'image1.jpg', display_id: 1 },
+            });
+        });
+        probe.mockResolvedValue({ width: 400, height: 300 });
+        probe.sync.mockReset();
 
         queries.managed.getBucket.mockResolvedValueOnce({
             row: { BucketName: 'my-bucket', Region: 'us-east-1', AccessKeyId: 'AKIA...', SecretAccessKey: 'secret' },
@@ -89,8 +109,8 @@ describe('getAnnotatePage - S3-backed image serving', () => {
 
         await getAnnotatePage(req, res);
 
-        expect(global.fs.readFileSync).not.toHaveBeenCalled();
-        expect(global.probe.sync).not.toHaveBeenCalled();
+        expect(fs.readFileSync).not.toHaveBeenCalled();
+        expect(probe.sync).not.toHaveBeenCalled();
         expect(s3Client.getObjectStream).toHaveBeenCalledWith(
             { fakeClient: true },
             'my-bucket',
@@ -106,21 +126,17 @@ describe('getAnnotatePage - S3-backed image serving', () => {
     it('still reads a locally-present file straight from disk, unaffected by the S3 changes', async () => {
         const imageRow = { IName: 'image1.jpg', reviewImage: 0, Source: null, SourceKey: null };
 
-        global.fs = {
-            existsSync: jest.fn().mockReturnValue(true),
-            readFileSync: jest.fn().mockReturnValue(Buffer.from('img-bytes')),
-        };
-        global.sqlite3 = {
-            Database: jest.fn((dbPath, cb) => {
-                cb && cb(null);
-                return makeFakeProjectDb({
-                    imagesRows: [imageRow],
-                    displayRow: { IName: 'image1.jpg', display_id: 1 },
-                });
-            }),
-        };
-        global.probe = jest.fn();
-        global.probe.sync = jest.fn().mockReturnValue({ width: 800, height: 600 });
+        fs.existsSync.mockReturnValue(true);
+        fs.readFileSync.mockReturnValue(Buffer.from('img-bytes'));
+        sqlite3.Database.mockImplementation((dbPath, cb) => {
+            cb && cb(null);
+            return makeFakeProjectDb({
+                imagesRows: [imageRow],
+                displayRow: { IName: 'image1.jpg', display_id: 1 },
+            });
+        });
+        probe.mockReset();
+        probe.sync.mockReturnValue({ width: 800, height: 600 });
 
         await getAnnotatePage(req, res);
 
@@ -136,16 +152,15 @@ describe('getAnnotatePage - S3-backed image serving', () => {
     it('renders 404 when there is no local file and the image is not S3-backed', async () => {
         const imageRow = { IName: 'image1.jpg', reviewImage: 0, Source: null, SourceKey: null };
 
-        global.fs = { existsSync: jest.fn().mockReturnValue(false), readFileSync: jest.fn() };
-        global.sqlite3 = {
-            Database: jest.fn((dbPath, cb) => {
-                cb && cb(null);
-                return makeFakeProjectDb({
-                    imagesRows: [imageRow],
-                    displayRow: { IName: 'image1.jpg', display_id: 1 },
-                });
-            }),
-        };
+        fs.existsSync.mockReturnValue(false);
+        fs.readFileSync.mockReset();
+        sqlite3.Database.mockImplementation((dbPath, cb) => {
+            cb && cb(null);
+            return makeFakeProjectDb({
+                imagesRows: [imageRow],
+                displayRow: { IName: 'image1.jpg', display_id: 1 },
+            });
+        });
 
         await getAnnotatePage(req, res);
 

From 488d30991468475825caeb4fef087446961fb00f Mon Sep 17 00:00:00 2001
From: FerretCode <eggertet@oregonstate.edu>
Date: Thu, 20 Aug 2026 13:07:43 -0700
Subject: [PATCH 18/18] fix several bugs

---
 queries/projects/projects.js                  | 28 +++------
 routes/pages/getAnnotatePage.js               |  8 +--
 tests/integration/legacyDbMigration.test.js   | 57 +++++++------------
 tests/unit/getAnnotatePageS3Streaming.test.js | 20 ++++---
 4 files changed, 46 insertions(+), 67 deletions(-)

diff --git a/queries/projects/projects.js b/queries/projects/projects.js
index 3591da47..9c82f46e 100644
--- a/queries/projects/projects.js
+++ b/queries/projects/projects.js
@@ -204,20 +204,6 @@ module.exports = {
             await db.run(
                 "CREATE TABLE IF NOT EXISTS Images (IName VARCHAR NOT NULL PRIMARY KEY, reviewImage INTEGER NOT NULL DEFAULT 0, validateImage INTEGER NOT NULL DEFAULT 0, Source VARCHAR DEFAULT NULL, SourceKey VARCHAR DEFAULT NULL)",
             );
-            try {
-                await db.run(
-                    "ALTER TABLE Images ADD COLUMN reviewImage INTEGER NOT NULL DEFAULT 0",
-                );
-            } catch (e) {
-                // Column already exists
-            }
-            try {
-                await db.run(
-                    "ALTER TABLE Images ADD COLUMN validateImage INTEGER NOT NULL DEFAULT 0",
-                );
-            } catch (e) {
-                // Column already exists
-            }
             await db.run(
                 "CREATE TABLE IF NOT EXISTS Labels (LID INTEGER PRIMARY KEY, CName VARCHAR NOT NULL, X VARCHAR NOT NULL, Y VARCHAR NOT NULL, W INTEGER NOT NULL, H INTEGER NOT NULL, IName VARCHAR NOT NULL, FOREIGN KEY(CName) REFERENCES Classes(CName), FOREIGN KEY(IName) REFERENCES Images(IName))",
             );
@@ -225,18 +211,20 @@ module.exports = {
                 "CREATE TABLE IF NOT EXISTS Validation (Confidence INTEGER NOT NULL, LID INTEGER NOT NULL PRIMARY KEY, CName VARCHAR NOT NULL, IName VARCHAR NOT NULL, FOREIGN KEY(LID) REFERENCES Labels(LID), FOREIGN KEY(IName) REFERENCES Images(IName), FOREIGN KEY(CName) REFERENCES Classes(CName))",
             );
 
-            // Images predates the Source/SourceKey columns, so CREATE TABLE IF NOT EXISTS
-            // above is a no-op on any project database created before this change.
-            // Back-fill them here, guarded by a PRAGMA check since SQLite has no
-            // ADD COLUMN IF NOT EXISTS. SourceKey holds the literal S3 object key (which
-            // may differ from IName once collisions are disambiguated), decoupled from the
-            // display name.
+            // Images predates the reviewImage/validateImage/Source/SourceKey columns, so
+            // CREATE TABLE IF NOT EXISTS above is a no-op on any project database created
+            // before this change. Back-fill them here, guarded by a PRAGMA check since
+            // SQLite has no ADD COLUMN IF NOT EXISTS. SourceKey holds the literal S3 object
+            // key (which may differ from IName once collisions are disambiguated), decoupled
+            // from the display name.
             const imageColumns = await db.all("PRAGMA table_info(Images)");
             const existingColumnNames = new Set(
                 (imageColumns.rows || []).map((column) => column.name),
             );
 
             const backfillColumns = [
+                { name: "reviewImage", ddl: "reviewImage INTEGER NOT NULL DEFAULT 0" },
+                { name: "validateImage", ddl: "validateImage INTEGER NOT NULL DEFAULT 0" },
                 { name: "Source", ddl: "Source VARCHAR DEFAULT NULL" },
                 { name: "SourceKey", ddl: "SourceKey VARCHAR DEFAULT NULL" },
             ];
diff --git a/routes/pages/getAnnotatePage.js b/routes/pages/getAnnotatePage.js
index 7c81255a..cd694a77 100644
--- a/routes/pages/getAnnotatePage.js
+++ b/routes/pages/getAnnotatePage.js
@@ -1,11 +1,11 @@
-const path = require("path");
-const fs = require("fs");
-const sqlite3 = require("sqlite3").verbose();
-const probe = require("probe-image-size");
 const queries = require("../../queries/queries");
 const { buildS3Client, getObjectStream } = require("../../utils/s3Client");
 
 async function getAnnotatePage(req, res) {
+    var path = global.path || require("path");
+    var fs = global.fs || require("fs");
+    var sqlite3 = global.sqlite3 || require("sqlite3").verbose();
+    var probe = global.probe || require("probe-image-size");
     var IDX = parseInt(req.query.IDX),
         IName = String(req.query.IName),
         curr_class = req.query.curr_class,
diff --git a/tests/integration/legacyDbMigration.test.js b/tests/integration/legacyDbMigration.test.js
index 188ae6cf..a8e675ab 100644
--- a/tests/integration/legacyDbMigration.test.js
+++ b/tests/integration/legacyDbMigration.test.js
@@ -1,6 +1,6 @@
 const path = require('path');
 const fs = require('fs');
-const sqlite3 = require('sqlite3').verbose();
+const { Client } = require('../../queries/client');
 const queries = require('../../queries/queries');
 
 describe('Legacy Database Migration for Images validateImage / reviewImage columns', () => {
@@ -14,7 +14,10 @@ describe('Legacy Database Migration for Images validateImage / reviewImage colum
             fs.mkdirSync(tmpDir, { recursive: true });
         }
         dbPath = path.join(tmpDir, 'test_legacy.db');
-        global.projectDbClients[dbPath] = new sqlite3.Database(dbPath);
+        // Mirror how production populates global.projectDbClients (see server.js), so
+        // migrateProjectDb's promise-based db.run/db.all calls behave the same way they
+        // do outside tests, instead of hitting the raw sqlite3.Database callback API.
+        global.projectDbClients[dbPath] = new Client(dbPath);
     });
 
     afterEach(() => {
@@ -26,49 +29,33 @@ describe('Legacy Database Migration for Images validateImage / reviewImage colum
 
     test('migrateProjectDb adds missing reviewImage and validateImage columns to existing Images table', async () => {
         const db = global.projectDbClients[dbPath];
-        db.run('CREATE TABLE Images (IName VARCHAR NOT NULL PRIMARY KEY)');
-        db.run("INSERT INTO Images (IName) VALUES ('test_image.jpg')");
+        await db.run('CREATE TABLE Images (IName VARCHAR NOT NULL PRIMARY KEY)');
+        await db.run("INSERT INTO Images (IName) VALUES ('test_image.jpg')");
 
         // Run migrateProjectDb
         await queries.project.migrateProjectDb(dbPath);
 
         // Verify query succeeds
-        await new Promise((resolve, reject) => {
-            const pdb = global.projectDbClients[dbPath];
-            pdb.all(
-                'SELECT Images.IName, Images.reviewImage, Images.validateImage FROM Images',
-                [],
-                (err, rows) => {
-                    if (err) return reject(err);
-                    expect(rows).toBeDefined();
-                    resolve();
-                }
-            );
-        });
+        const result = await db.all(
+            'SELECT Images.IName, Images.reviewImage, Images.validateImage FROM Images',
+        );
+        expect(result.rows).toBeDefined();
     });
 
     test('projectsFilter route handler query auto-migrates missing columns', async () => {
         const db = global.projectDbClients[dbPath];
-        db.run('CREATE TABLE Images (IName VARCHAR NOT NULL PRIMARY KEY)');
+        await db.run('CREATE TABLE Images (IName VARCHAR NOT NULL PRIMARY KEY)');
 
         // Simulate reading legacy db using the same logic as getProjectPage / projectsFilter
-        await new Promise((resolve, reject) => {
-            const pdb = global.projectDbClients[dbPath];
-            pdb.run("ALTER TABLE Images ADD COLUMN reviewImage INTEGER NOT NULL DEFAULT 0", () => {
-                pdb.run("ALTER TABLE Images ADD COLUMN validateImage INTEGER NOT NULL DEFAULT 0", () => {
-                    const query = `
-                        SELECT Images.IName, Images.reviewImage, Images.validateImage, COUNT(Labels.LID) AS numLabels
-                        FROM Images
-                        LEFT JOIN Labels ON Images.IName = Labels.IName
-                        GROUP BY Images.IName
-                    `;
-                    pdb.all(query, [], (err, rows) => {
-                        if (err) return reject(err);
-                        expect(rows).toBeDefined();
-                        resolve();
-                    });
-                });
-            });
-        });
+        await db.run("ALTER TABLE Images ADD COLUMN reviewImage INTEGER NOT NULL DEFAULT 0");
+        await db.run("ALTER TABLE Images ADD COLUMN validateImage INTEGER NOT NULL DEFAULT 0");
+        const query = `
+            SELECT Images.IName, Images.reviewImage, Images.validateImage, COUNT(Labels.LID) AS numLabels
+            FROM Images
+            LEFT JOIN Labels ON Images.IName = Labels.IName
+            GROUP BY Images.IName
+        `;
+        const result = await db.all(query);
+        expect(result.rows).toBeDefined();
     });
 });
diff --git a/tests/unit/getAnnotatePageS3Streaming.test.js b/tests/unit/getAnnotatePageS3Streaming.test.js
index a1f3931f..ead168b9 100644
--- a/tests/unit/getAnnotatePageS3Streaming.test.js
+++ b/tests/unit/getAnnotatePageS3Streaming.test.js
@@ -24,17 +24,21 @@ const getAnnotatePage = require('../../routes/pages/getAnnotatePage');
 // Mirrors the real sqlite3 driver's callback style, since getAnnotatePage.js
 // wraps `this.get`/`this.all` in its own Promise-returning getAsync/allAsync
 // right after construction - only the raw callback methods are ever called.
+// getAsync/allAsync always invoke the 3-arg form (sql, params, callback), so
+// these must accept that arity rather than the 2-arg (sql, callback) form.
 function makeFakeProjectDb({ classesRows = [], labelsRows = [], imagesRows = [], displayRow }) {
     return {
-        get: jest.fn((sql, cb) => {
-            if (sql.includes('display_id')) return cb(null, displayRow);
-            return cb(null, undefined);
+        get: jest.fn((sql, params, cb) => {
+            const callback = typeof params === 'function' ? params : cb;
+            if (sql.includes('display_id')) return callback(null, displayRow);
+            return callback(null, undefined);
         }),
-        all: jest.fn((sql, cb) => {
-            if (sql.includes('Classes')) return cb(null, classesRows);
-            if (sql.includes('Labels')) return cb(null, labelsRows);
-            if (sql.includes('Images')) return cb(null, imagesRows);
-            return cb(null, []);
+        all: jest.fn((sql, params, cb) => {
+            const callback = typeof params === 'function' ? params : cb;
+            if (sql.includes('Classes')) return callback(null, classesRows);
+            if (sql.includes('Labels')) return callback(null, labelsRows);
+            if (sql.includes('Images')) return callback(null, imagesRows);
+            return callback(null, []);
         }),
         each: jest.fn((sql, cb) => cb(null, undefined)),
         close: jest.fn((cb) => cb && cb(null)),