This project demonstrates browser-based proctoring using TensorFlow.js models for face and object detection. Session events and recordings are saved to Firebase for later review.
- Detects faces to track user attention and triggers events when the user looks away or multiple faces appear.
- Uses COCO-SSD to flag suspicious objects such as phones, books or laptops.
- Records the camera stream and uploads the video to Firebase Storage.
- Logs session metadata and events in Firestore for persistence.
- Navigate into the app directory and install dependencies:
cd focus-proctor npm install - The Firebase project configuration lives in
src/firebase.ts. Replace the values with your own configuration if needed.
Run a local dev server:
npm run devCompile TypeScript and bundle the app with Vite:
npm run buildServe the production build locally:
npm run preview