Haar-VGG: Face Attendance System

Abstract

Attendance taking is a crucial practice in educational institutions in Malaysia, but the traditional manual method is time-consuming and risky, particularly in the post-Covid era. To address this, a face recognition attendance system using Python is developed. The Viola-Jones algorithm known as Haar is utilized for face detection, and transfer learning on VGGFace is applied for model training, using 195 images from FLW dataset and volunteers among students. The system achieves a validation and testing accuracy of 1.0 through image preprocessing and augmentation. The attendance system includes a user-friendly graphical interface and live webcam feed, enabling instant recognition and recording of attendance. Integration with a MySQL database allows easy access to attendance records for teachers. This advanced system saves time, reduces the risk of virus transmission, and simplifies attendance management, offering a convenient and efficient solution for educational institutions.