Design And Implementation Of The Face Recognition Classroom Sign-in System Based On MTCNN And FaceNet

With the continuous development of the information age,machine learning has been widely used,and deep learning,as one of the important research directions,has developed rapidly in the field of face detection and face recognition.Traditional face detection and face recognition algorithms are easily affected by external environmental factors,which makes it difficult to collect face images and complete face recognition.Research on the light intensity,partial occlusion,and face pose rotation of face detection in the existing algorithms is relatively few,and at the same time,the accuracy of real-time face detection and recognition is reduced.Therefore,under the premise of considering the change of external factors,it is the direction of this paper to quickly and accurately recognize the face.In this paper,based on the convolutional neural network model,for the face pose rotation,partial occlusion and light intensity,the MTCNN model is proposed for face detection,and the FaceNet face model is used for face recognition;finally,based on the existing algorithm,Completed the design and implementation of the face recognition sign-in system.The main work of this article is as follows:(1)According to a series of changing factors in the detection process,a multi-cascade CNN model is proposed.For real-time detection of the face in the camera,the multi-layer convolutional neural network predicts the face layer by layer.Through the MTCNN face detection method to obtain the position of the face image and the position of the key points of the face,the obtained face image is cropped and normalized to reduce the impact of environmental factors such as lighting on the accuracy of face recognition.(2)Use the FaceNet face recognition model to extract features from face images,and obtain single recognition results through SVM.Based on the FaceNet deep model,the Inception-resnet-V1 network structure is used,the Triplet Loss ternary loss function is selected,and the LFW face data set is used to test the FaceNet deep CNN model.(3)Based on the existing algorithm,combined with the camera and Sqlite database,using B / S architecture,the face recognition classroom sign-in system is designed andimplemented.The database stores face information,including student ID,name,gender,major,and face characteristics.