Design And Implementation Of Employee Identity Verification System Based On Face Recognition

With the continuous progress and development of human society,people's lives have undergone tremendous changes,and the flow of humans has accelerated.For the management of personnel,especially in the identity verification process,government,enterprises,schools and other gathering places need to be efficient,Fast and safe verification certification.Therefore,what method is used for identity verification becomes the key.Although traditional fingerprint recognition is fast,it has low security and cannot meet the requirements of non-contact verification.As deep learning shines,especially in the field of face recognition,its powerful feature extraction and excellent classification capabilities are very robust to face verification,and improve the accuracy and safety of recognition.The following is the main research done in this paper on the application of deep learning to the identification of employees in small and medium-sized companies.(1)In the first core stage of identity verification,that is,face detection,this paper studies the use of traditional haar feature detection and Dlib-based convolutional neural network detection algorithms,using two algorithms to crop the input of face images Wait for the preprocessing operation,and then use the CASIA-Face V5 data set to experiment with it,compare the detection accuracy,perform static comparison of the front and side faces in the selected test image,and perform dynamic face detection based on the person in the actual environment.Compare and comprehensively analyze the detection capabilities of the two.(2)By comparing the research and experiment of face detection,on the basis of detection,the Keras framework is used to build and design the convolutional neural network structure,the loss function selects the cross-entropy loss,and the operations of generator training and data promotion introduced in the framework are combined with haar feature detection.As the first authentication algorithm.Use the residual network to add a fully connected layer at the end to extract face features,and combine it with thesecond detection method Convolutional Neural Network as the second verification algorithm.Use the two identification schemes to perform the same training and testing on the MORPH data set and the comparison of the identity verification test in the actual environment,and select the better scheme as the identification algorithm of the identification verification system.(3)Finally,based on the above research scheme,this article develops an authentication system using Python as the back-end development language,and the interface side is designed using Python third-party library Pyqt.The system design and system function modules are introduced in detail,and the actual environment tests of face detection and face recognition on this system have achieved good results and also proved the feasibility of this system in identity verification.