software projectCompleted
True Prence
Face recognition attendance system identifying up to 20 students per frame simultaneously. CNN (GPU) or HOG (CPU) detection, SQLite storage, Tkinter UI, Excel/PDF/chart exports.
- Faces per frame
- Up to 20
- Detection
- CNN + HOG
- GPU
- CUDA / dlib
Where it started
Manual attendance tracking in classrooms and workplaces is slow, error-prone, and easily gamed by proxy attendance.
How I built it
Built a single-file face recognition system using dlib CNN/HOG detection, processing 20 faces per frame with automatic GPU detection, full SQLite data model, and export pipelines.
What it makes possible
Enables passive, zero-interaction attendance for full classrooms — students are marked present simply by being visible to the camera.
Built with
- AI
- Python
- OpenCV
- face_recognition
- SQLite