Attendance, by recognition.
Register students once. Train a small CNN. Attendance records itself every time the webcam sees a familiar face.
How it works
- Enroll a student. Open the Enroll tab, enter their ID and name, and click Capture. 30 face crops are saved automatically.
- Train the CNN. One click. In under a minute the model learns to tell your enrolled students apart.
- Recognize & mark. The Recognize tab watches the webcam and writes attendance.csv as it sees familiar faces.
Recent attendance
Enroll a new face.
Face the camera, name the student, and the system records 30 grayscale face crops. Move your head gently for variety.
Train the CNN.
Teaches a small convolutional network to tell your enrolled students apart. Needs at least two students. Takes under a minute on most laptops.
Click Train now. The server runs src/train_cnn.py, saves the model and reloads it for recognition.
—
Live recognition.
The webcam streams through the CNN. Known students are marked present (one entry per student per day). Unknown faces are left alone.
The attendance log.
Every recognised student lands here with their date, time and status. Only one entry per student per day.