Facial recognition · classroom ready

Smart attendance for modern campuses

Enroll students with a few photos, extract MobileNetV2 embeddings, and mark attendance from any device with a camera—verified by cosine similarity and your institution’s rules.

Open student portal Mark attendance How it works
Role-based portals Separate flows for students, faculty, and administrators.
Reports & exports Attendance grids, defaulters, CSV/XLSX where configured.
Notices College-wide and department messaging for students.

Choose an entry point

Students

Sign in with roll number or portal username to view today’s marks, subject-wise stats, notices, and downloadable reports.

Student sign in

Quick mark

No login required—capture your face when prompted. Recognition matches against enrolled embeddings only.

Open camera flow

Faculty

Scoped to assigned subjects and sections: manual attendance, corrections, face capture, and class analytics.

Faculty login

Administration

Roster and face dataset, academic structure, faculty assignments, system thresholds, and global exports.

Admin login
Demo notice

Deployments should use HTTPS for camera APIs, set a strong SECRET_KEY, and treat biometric data according to your institution’s privacy policy.

Recognition pipeline
1
Detect OpenCV Haar cascade locates a frontal face in the frame.
2
Embed CNN (MobileNetV2 + Global Average Pooling) produces a fixed-length vector.
3
Match Cosine similarity vs. stored mean embedding; threshold from system settings.
4
Record Attendance row with time, optional snapshot path, and duplicate-window checks.
Technology snapshot
  • Flask · SQLAlchemy · SQLite
  • OpenCV · TensorFlow / Keras
  • Bootstrap 5 · responsive UI
  • Optional CSV / spreadsheet exports

Student portal dashboard · attendance · notices · profile