RESEARCH AND DEVELOPMENT OF A FACE RECOGNITION MODEL FOR STUDENT ROLL CALL APPLICATION

Abstract

Face recognition is one of the important research directions of computer vision, aiming to automatically identify or verify personal identity based on biometric features from facial images or videos. This paper presents a face recognition system applied to student attendance in the classroom. In this paper, we propose a MFSN system with three main stages including MTCNN detection, feature extraction using FaceNet and identity verification using one-shot learning mechanism using Siamese Network. Experimental results show that our proposed model achieves 98.3% accuracy on VGGFace2 and achieves high performance on a real dataset of HUIT students with only one registered image for each student. The system is fully deployed on a low-end personal computer, operating independently in a local area network (LAN) environment without requiring an Internet connection or cloud services. The system can take real-time attendance for a class of 50 students in an average time of 8 seconds