AI Attendance & Class Mgmt
Face-recognition attendance and class management for university classrooms — a MediaPipe + ArcFace pipeline wired into a MERN dashboard.
Manual roll-calls eat 10–15 minutes of every lecture, proxy attendance is trivial, and by exam season nobody trusts the registers. Faculty needed attendance that is fast, fraud-proof and auditable — without buying dedicated hardware.
I built the full pipeline: MediaPipe handles real-time face detection on a commodity webcam, ArcFace embeddings match identities against enrolled students, and a threshold-tuned matcher rejects lookalikes and photos. Recognition events stream into an Express/MongoDB API, and a Next.js dashboard gives faculty live class views, overrides and exportable reports — with role-based access for admins, faculty and students.
A full class is marked in under two minutes with no hardware beyond the room's webcam. Attendance disputes dropped to near zero because every mark carries a timestamped recognition record, and coordinators finally get semester-level analytics instead of paper registers.
“Hadi delivered a system our department can actually run — accurate, fast, and polished far beyond a typical final-year project.”