Hadirin

Flutter frontend with Firebase Auth for sessions and Firestore as the primary database. Face enrollment extracts a 128-d embedding on-device via MobileFaceNet and uploads the photo to Cloudinary, storing the vector in Firestore. At check-in, a live frame is processed through Google ML Kit for detection, then matched against the stored embedding using cosine similarity. Geo-fence validation runs client-side before the attendance record is committed to Firestore.