welcome to our final project

Project Members	Department
C21147354M
C21145681W
C21147799W
C21145989W
C21147067J
C21145793S
C21146707O
C21146057T

Overview

This repository contains research artifacts and a prototype Human-Machine Interface (HMI) for a computer vision system codenamed "Uzziah". The HMI is a Flask + Socket.IO app that displays a live video feed and overlays tracking/prediction results streamed from a backend detection/prediction pipeline.

Key components:

• Human Machine Interface (HMI): Flask server with a web UI, receives processed frames over Socket.IO and renders them in the browser.
• Detection/Prediction Engines: Background threads that read camera frames, run object detection + tracking, and emit processed frames to the HMI.
• Research materials: Design docs, tutorials, and resources used during development.

Repository Structure

uzziah/
├─ projectproposal.pdf
├─ ReadMe.md  (this file)
└─ research/
   ├─ algorithm design/
   ├─ documentation/
   ├─ hardware design/
   │  └─ sketches/
   └─ sofware design/
      ├─ Detection Engine/
      │  ├─ README.md
      │  └─ requirements.txt
      ├─ Human Machine Interface/
      │  ├─ app.py
      │  ├─ requirements.txt
      │  ├─ templates/
      │  │  └─ index.html
      │  └─ static/
      │     ├─ style.css
      │     └─ script.js
      │  └─ utils/
      │     ├─ detection_engine.py
      │     └─ prediction_engine.py
      ├─ ReadMe.MD
      ├─ resources/
      │  └─ weights.pt
      └─ tutorials/
         ├─ SORT.ipynb
         ├─ fronted_streaming_connection.ipynb
         └─ how_to_track_objects_with_sort_tracker.ipynb

HMI (Flask) — Quick Start (Windows PowerShell)

1. Navigate to the HMI directory:

   cd "research/sofware design/Human Machine Interface"

2. Create and activate a virtual environment (recommended):

   python -m venv .venv
   .\.venv\Scripts\Activate.ps1

3. Install dependencies:

   python -m pip install -r requirements.txt

4. Run the app:

   python app.py

5. Open the UI in your browser at:

   • http://localhost:5000

The UI shows a tactical HUD-style page. A background thread starts the detection engine and another thread emits frames to the frontend via Socket.IO. Use the "NEUTRALIZE" button to test a POST action to /neutralize.

Detection & Prediction Engines (Threads)

• Entry point: research/sofware design/Human Machine Interface/app.py

  • Spawns two daemon threads:
    • utils.detection_engine.detection_engine(frame_queue, result_queue)
    • utils.prediction_engine.prediction_engine(result_queue, hmi_socket)

• Detection engine (utils/detection_engine.py):

  • Reads frames from a camera, runs YOLO for detection, then SORT for tracking.
  • Publishes (frame, tracks) tuples into result_queue.

• Prediction engine (utils/prediction_engine.py):

  • Consumes (frame, tracks), applies overlays, encodes frames as JPEG base64.
  • Emits frames to the web client via Socket.IO (video_frame event).

Dependencies

HMI (research/sofware design/Human Machine Interface/requirements.txt):

• Flask, Flask-SocketIO, OpenCV, inference-gpu, trackers, supervision==0.27.0rc1

Detection Engine (research/sofware design/Detection Engine/requirements.txt):

• ultralytics (YOLO) and a SORTTracker implementation are imported in utils/detection_engine.py but are not listed in the HMI requirements. You may need to install them manually:

  python -m pip install ultralytics

Development Tips

• If Socket.IO events are not reaching the browser, verify the app is started via SocketIO.run (already configured in app.py) and that the client loads the correct Socket.IO script.
• Check the console output for background thread exceptions.

Contributing

• Open issues or PRs for improvements.