Real-time AI that detects student disengagement via webcam and delivers personalized nudges — keeping every student on track while giving teachers live classroom insights.
- CV Pipeline: Python, YOLOv8n, OpenCV
- Backend: Python, FastAPI, Deepgram (live transcription), Claude API (AI nudges), Supabase
- Frontend: Next.js 16, React 18, Tailwind CSS
- Infra: WebSockets
The system has four components that work together in real time:
- CV Pipeline — Captures webcam frames, detects head pose, gaze, and posture to identify disengaged students
- Backend — Receives disengagement events, pulls transcript context via Deepgram, and calls Claude to generate a personalized nudge
- Student UI — Displays nudges as chat messages the student receives on their device
- Teacher Dashboard — Live heatmap of class engagement, transcript timeline, and per-student chat logs
- Python 3.11+
- Node.js 18+
# Clone the repo
git clone https://github.com/Samuaray/ClassAssist.AI.git
cd ClassAssist.AI/clarityai
# Copy and fill in environment variables
cp .env.example .env