How innocentmchry/studentmonitoring works
Project type and purpose
• This is a full-stack, real-time online classroom monitoring system built on Django (Python) with a WebRTC-based video conferencing frontend and an integrated CNN emotion-detection pipeline. It supports live emotion tracking of students (via Agora’s WebRTC SDK) and aggregates session summaries.
File structure
• Top-level:
• manage.py, gunicorn.conf.py, requirements.txt → Django entry point & deployment
• mychat/ → Django project settings, ASGI/WSGI, URL routing
• base/ → Django “app” containing models, views, forms, templates, URL patterns, tests, migrations
• static/ → client-side assets (JS, CSS, AgoraRTC library, images)
• savedModels/ → pretrained CNN model files, Haarcascade XML for face detection
• ssl/ → self-signed SSL certificate & key
• Indicators:
• asgi.py implies support for async/WebSocket (though actual realtime video is via Agora)
• static/js/streams.js & summary.js handle client-side WebRTC logic and summary fetching
• base/views.py likely exposes REST endpoints for model inference & summary data
• db.sqlite3 for local development; recommend PostgreSQL in prod
• gunicorn.conf.py for production WSGI server
README for additional insights
• Architecture image (link) describes:
• Browser clients ↔ Agora Real-Time Messaging/RTC service for video
• Backend Python service: face detection + CNN inference → emotion valence/arousal → map to one of five states
• Data persistence & summary generation
• Tech stack: Django, WebRTC (via AgoraRTC_N-4.19.1.js), Keras/TensorFlow model (.h5 + .json), Haarcascade
Details
• Real-time video flow vs. periodic inference/summarization flow distinctly
• Where SSL/TLS is terminated and how Gunicorn/Nginx fit in