Entrance Guard Design Based On Embedded Face Recognition

In the face of the rapid development of society,people’s pace of life has become faster and faster.The use of traditional access control systems such as magnetic cards or keys has exposed problems such as inability to authenticate and easy to lose,which has been unable to meet people’s needs.With the advent of the intelligent era,the use of biometric identification technology for identity authentication has become the mainstream.This paper aims to design an intelligent access control system to achieve the purpose of rapid and accurate identification of identity.A face recognition access control system based on embedded structure is designed.With raspberry pie development board as the core,the algorithm is transplanted to raspberry pie development board,and then the hardware platform of the access control system is built.According to the actual needs,the interface of access control system is designed.Facedetection and recognition algorithm is the main research content of the access control system in this paper.Through experimental analysis and comparison of Ada Boost algorithm and MTCNN algorithm,MTCNN algorithm is selected as the face detection algorithm of the system.FaceNet algorithm is used for face recognition,and the network is lightweighted.Mobile Net model is selected for the backbone network.The MTCNN algorithm is used to detect the face target and extract the face candidate region.The FaceNet algorithm extracts the feature vector and compares the faces in the database.The self-built face dataset is trained on the pre-trained FaceNet model,and the accuracy of the trained network reaches 97.80 %.The face recognition accuracy of the Mobile Net model and the original model is tested by the test set.The test results show that the accuracy is improved by 1.10 %.The system performance is tested in the practical application scenario,and the factors affecting the performance are analyzed.The experimental results show that the designed system can meet the expected standards of practical application scenarios in face detection rate,face recognition rate and time-consuming,and the designed human-computer interaction interface can meet daily needs.The system has excellent design performance and good application prospect.