The Student Attendance Control System Based on Face Recognition

Abstract

Currently, one of the significant factors for improving the quality of training of specialists is the control of student attendance. This process can be automated. The paper suggests an approach to building a student attendance control system based on face recognition technology, which allows you to identify many people at the same time without direct contact with them and without using expensive equipment. This approach is based on the convolutional neural networks RetinaFace and ResNet, selected based on the review of modern methods of facial recognition presented in the paper. The architecture of our attendance control system is complemented by image preprocessing procedures, which, according to our proposed method based on the BREN measure, check the image quality and, if necessary, apply algorithms to the image to reduce noise, sharpen, increase brightness and align colors. The results of computational experiments are presented, which have shown a higher efficiency of the proposed approach compared with analogues.