Research And Implementation Of An Identity Verification System Based On Deep Learning

The normal operation of society is inseparable from the identification of people's identity.With the acceleration of social rhythm,it is the general trend to complete the comparison of people's identity through machine instead of manual in places such as airports,stations,hotels and so on,which need extensive personnel and certificate verification.With the development of deep learning technology,face recognition technology based on convolutional neural network has shown great advantages over traditional algorithms in relevant applications.Many excellent face recognition algorithms are still refreshing the data of various professional competitions.Applying face detection and face recognition algorithm based on deep learning technology to the identity verification system,further improving the efficiency of the identity verification system is a subject worthy of study.At the same time,the effect of face recognition in the application scene is usually affected by many factors,such as lighting conditions,occlusion,face posture,etc.How to enhance the robustness of face recognition algorithm,how to detect real-time face quickly and accurately,how to optimize the algorithm to improve the efficiency of identity verification system,all need to carry out systematic research and analysis.In this paper,the construction of a person to identity verification system based on deep learning is realized,and the implementation and improvement of face detection and face recognition algorithms are mainly studiedIn this paper,the construction of a person to identity verification system based on deep learning is realized,with the emphasis on the implementation and improvement of face detection and face recognition algorithms.A big problem in the scene of identity verification is how to recognize the face quickly and accurately.Based on the realization of the face detection algorithm MTCNN network,combined with the related knowledge of object detection,this paper improves the face detection algorithm and proposes the improved MTCNN algorithm based on Kalman filter,which significantly improves the performance of the face detection algorithm.In the aspect of face recognition algorithm design,through training and testing the classic face recognition network Facenet,this paper proposes an improved face feature extraction network,which not only meets the requirements of accuracy,but also speeds up the model training speed and convergence time.The original concept Inception-Res Net-v4 model is lightweight to reduce the parameters of the feature extraction network.In the aspect of functional innovation of identity verification system,based on the realization of the basic face comparison function,this paper develops the staff mode functionof the system,and achieves the unification of the function of face comparison and face recognition.This paper designs an identity verification system based on deep learning,including face acquisition module,ID card information acquisition module,face detection module,face comparison module and user interface module,and evaluate the function and performance of the system through a series of experiments.Through the experimental test,the accuracy of the system designed in this paper can reach 96%,which can better meet the application requirements.The additional function employee mode of the system can be realized successfully,which further increases the practicability of the system.