Based On The Improved DEEPID2 Integration Of Person And Certificate And The Examination Room Identity Verification System

In today's society where image recognition is widely used,due to its separability from people,traditional identity verification methods such as IC cards are prone to embezzlement or imitation.Therefore,in the important transportation hubs between cities such as railway stations and airports,the technology of integration of person and certificate replaces the traditional verification methods.In daily life,however,some large,authentication is required,visitors in examination room scenes,for example,often still rely on manpower to tested one by one,which not only increased the labor intensity of the examiners,also hard to avoid can appear outside the phenomenon such as congestion and human error,which in turn leads to factors that influence the equity of examination,and the witness and technology in order to guarantee the safety of some important areas and check for a long time,it is difficult to deal with short-term traffic scene.Therefore,this paper focuses on the identity verification in the examination room,and designs a system based on the improved DEEPID2,which collects the information of the examinee by the camera and the card reader respectively,so as to carry out fast and accurate identity verification.The main research contents of this paper are as follows:At First,by analyzing the changes of the DEEPID third-generation network and the convolutional neural network,the structure and loss function combination of the DEEPID 2 network are adjusted from different aspects,and the comparative experiments are carried out to find out the factors that affect the classification accuracy of the DEEPID 2 network,and then the improvement scheme of the DEEPID 2 network is determined.Secondly,Then,the improved DEEPID2 network is used for face verification experiments,and the network test set and the self-built actual test set are used to test the improved DEEPID2 network.The experimental results show that the improved DEEPID2 network has a higher accuracy rate of 99.51% than the original network.Finally,according to the demand analysis of the actual examination room,the design of the examination room identity verification system,and the improved DEEPID2 network combined with the system,to realize the examination room identity authentication,at the same time feedback results in the interface.After the test,the verification process takes about 700 milliseconds in the system,and the operation time of the whole system is about 2 seconds,which can meet the short-term identification requirements in the actual test room.