Mobile Attendance System Based On Face Recognition

Deep learning technology were used to achieve live detection and face recognition,and develop a face recognition sign-in system.The system function is implemented by the face sign-in App deployed on the mobile terminal.The teacher opens the mobile phone camera to record the face of the classmate for attendance,and manage courses and student information conveniently on the mobile terminal.In this paper,the gray scale transformation and the maximum between-class variance method are selected to normalize the gray level of the face image with experimental comparison and analysis,and the face detection algorithm based on structural features is used to achieve the geometric normalization of the face image;it is determined by the accuracy rate comparison Use the SPXY algorithm to divide the sample set.Based on the deep learning frameworks Keras and Tensor Flow,two models are trained:based on the difference between living and prostheses,comparing the training effects of different loss functions of the livenessnet convolutional network for training the living detection model;based on facial feature information,comparing the Inception V3 neural network Different loss calculation methods for training face recognition models.The mobile app is deployed on an Android phone,the classmates are "swiping their faces" with the camera.After the App detects the live facial features,it sends query information to the database,and the back-end will send real-time facial data and related attendance The information returns to the App interface.The system simplifies the work process of class sign-in and statistics of attendance scores,significantly speeds up the speed,avoids the phenomenon of signing substitutes,and completes the integration of the collection and management of student attendance information.