2025–2026
Smart Attendance (Final Year Project)
Face-recognition attendance system with a REST API, admin dashboard and exportable reports.
PythonFastAPIOpenCVReactDocker
The problem
Manual roll calls wasted around ten minutes per lecture and were easy to fake.
What I built
Trained a face-embedding pipeline with OpenCV and served it behind a FastAPI service. A React dashboard lets lecturers review and correct matches.
Containerised everything with Docker Compose so the whole stack starts with one command.
Results
96% match accuracy on a 120-student test set; attendance now takes under a minute.
What I learned
How to evaluate a model honestly with a held-out set, and why privacy and consent need to be designed in from the start.