Attendance System Based on Dynamic Face Recognition
Shizhen Huang, Haonan Luo · 2020
A video-based attendance system is designed by using the method of real-time face recognition. The system supports multi-user attendance and face liveness detection at the same time. The system can automatically collect face data, which will be saved in the database as well as attendance results. The face detection part of the system is based on MTCNN (Multitask Convolutional Neural Network) algorithm, and the face recognition part is based on FaceNet algorithm. The algorithm implementation is based on TensorFlow framework, and the face liveness detection part is based on ERT (Ensemble of Regression Tree) algorithm, which can judge whether the user blinks. The attendance system is written in Python language, and the user interface is designed by Qt library. The experimental results show that the system achieves a good performance in real-time face recognition. The false accept rate and false rejection rate of face recognition are within 2%, and the recognition rate can be stable at 20 FPS.