Implementation of a Facial Recognition Attendance System for UNIZIK Department of Computer Engineering

Proxy attendance and inaccurate manual records in Nigerian university lecture halls undermine attendance policy enforcement and impede student performance monitoring. This study implements a facial recognition-based attendance system for the Department of Computer Engineering at Nnamdi Azikiwe University (UNIZIK), Awka, using computer vision to automatically identify and log student presence at the start of each lecture. The system is built on a Raspberry Pi 4 with a Pi Camera Module v2, using the face_recognition Python library based on the dlib deep learning framework for face detection and recognition. An SQLite database stores enrolled face encodings and attendance logs, with a web-based dashboard accessible to lecturers for attendance review and reporting. Student enrollment was conducted via a controlled photo capture session. System evaluation was carried out across five practical sessions in the department, involving 87 enrolled students. Results indicate a face recognition accuracy of 93.8 percent under standard indoor lighting conditions, declining to 86.2 percent under poor illumination. False acceptance rate was 0.9 percent, and false rejection rate was 6.2 percent. Average identification time per student was 1.1 seconds. Lecturer satisfaction with the system interface averaged 4.3 out of 5. The study concludes that facial recognition attendance systems are technically viable for Nigerian university lecture environments with controlled lighting. Recommendations include infrared lighting enhancement for low-light conditions and integration with UNIZIK's central student records system.