Lecture Slide Capture

Local-first desktop app that watches a lecture video or browser window, saves only the moments when slides change, and exports the captured slides as images plus a slides.pdf.

Why This Exists

Lecture recordings are often long, repetitive, and difficult to review. Students usually need the slide states, not a full video recording. Lecture Slide Capture turns a live lecture window into a compact study artifact by detecting slide transitions locally and saving only meaningful frames.

The app is designed for privacy-sensitive study workflows:

• It runs on the user's machine.
• It does not upload lecture video or screenshots to a server.
• It stores slide images, logs, and the generated PDF in a local session folder.
• It works with a browser window or a manually selected screen region.

Who It Helps

• Students reviewing online lectures without manually screenshotting every slide.
• Educators and TAs who want quick slide snapshots from recorded lectures.
• Researchers building local-first note-taking and lecture-processing workflows.
• Accessibility-minded users who need a compact PDF view of visual lecture material.

Features

• Select a Chrome lecture window or a screen region from the GUI.
• Pick a slide ROI so browser chrome, captions, and controls can be excluded.
• Save images only when slide transitions are detected.
• Review the latest saved slide and recent saved slides while capture runs.
• Generate slides.pdf when the session ends.
• Pause, resume, and finish capture safely.
• Remember the output directory between sessions.
• Switch between English and Korean from the app language setting.
• Package a runnable macOS .app bundle and Windows build files.

Quick Start

macOS

1. Open Lecture Slide Capture.app.
2. If Python packages are missing, run the install command shown by the app.
3. Choose a Chrome lecture window or screen region.
4. Select the slide area.
5. Press Start Capture.
6. When finished, press Finish to generate slides.pdf.

macOS Screen Recording permission may be required. If capture returns a blank image, grant permission in System Settings and restart the app.

Windows

Build on Windows or run the GitHub Actions Build Windows workflow. PyInstaller cannot cross-compile a Windows executable from macOS.

.\scripts\build_windows.ps1

The distributable executable is created at:

dist\LectureSlideCapture.exe

Output

Each run creates a timestamped session folder under the selected base output directory. The base directory is editable in the GUI and saved locally.

Typical session contents:

slide_0001.png
slide_0002.png
slides.pdf
captures.csv
duplicates.csv
capture_source.json

Repository Contents

Lecture Slide Capture.app/
  Contents/Resources/slide_capture_gui.py   GUI frontend
  Contents/Resources/slide_capture.py       capture engine
  Contents/Resources/requirements.txt       Python dependencies
design/
  lecture-slide-capture-app-screenshot.png  App screenshot
packaging/windows/
  LectureSlideCapture.windows.spec          PyInstaller spec
scripts/
  build_windows.ps1                         Windows build script

Validation

The current app bundle has been checked with:

python3 -m py_compile "Lecture Slide Capture.app/Contents/Resources/slide_capture_gui.py" \
  "Lecture Slide Capture.app/Contents/Resources/slide_capture.py"
bash -n "Lecture Slide Capture.app/Contents/Resources/run_capture_in_terminal.sh"

End-to-end capture still depends on OS screen-recording permissions and an available lecture window.

Roadmap

• Improve Windows window-capture reliability and packaging documentation.
• Add small sample fixtures for slide-transition detection regressions.
• Add clearer first-run diagnostics for missing screen permissions.
• Improve keyboard navigation and accessibility in the Tkinter GUI.
• Publish downloadable release artifacts for macOS and Windows.

Contributing

Issues and pull requests are welcome. See CONTRIBUTING.md for development notes, privacy expectations, and suggested contributions.

License

MIT. See LICENSE.

Korean

한국어 설명은 README_ko.md를 참고해 주세요.