Research And Design Of Campus Face Recognition System Based On Deep Learning

In recent years,the campus safety accidents with students as the main body are on the rise,which highlights the problem that the campus internal safety supervision system is not perfect.Most campuses are using video surveillance system as a means of campus security supervision,but the technology has a strong dependence on the artificial,easy to fail to report,and eventually lead to the occurrence of misfortune.With the development of computer vision technology,face recognition technology has been significantly improved in recognition accuracy and speed.The application of face recognition technology in campus safety supervision system can not only save human and material resources,but also improve the intelligent degree of campus safety supervision system.Face recognition equipment has certain requirements for the application environment,and can only be installed in the main road where students flow intensively.In order to achieve all-round safety supervision for students in school,this paper designs a ZigBee positioning subsystem to assist face recognition technology to realize the recognition and positioning of students in school.In order to strengthen the safety supervision system and reduce the hidden danger of safety accidents,this paper,based on the Jilin Provincial Department of science and technology project,carries out the research and design of the campus face recognition system based on deep learning.The system is designed in three aspects as follows:1.This paper improves and trains the SSD face detection algorithm.The detection algorithm takes SSD recognition algorithm as the backbone network,which is used for data collection of the shallow part of the network.The deep part of the network adopts the fusion of the structure of the ASFF network to strengthen the mining and extraction of deep semantic information;in the face recognition part,face recognition algorithm based on facenet is adopted.By comparing the European distance between the feature vectors embedded of two face images,whether the European distance is less than the setting Set the threshold to determine whether it is the same person.2.Based on ZigBee technology,this paper studies and designs a set of ZigBee subsystem to capture the location information of students,to assist the face recognition subsystem to complete the positioning of students on campus.When the campus students wear terminal nodes to move within the reference point range,they can decode the returned location information and save the location information to the database.3.Through the research and design of face recognition algorithm,hardware circuit and software function,this paper designs a campus face recognition system with face input function,face recognition function and tracking students' action trajectory function.The action track module of the system can judge the action track of students in a day through the face recognition positioning information,ZigBee positioning information and big data information of students in the school stored in the database.The action track module can be used to view the position status of students at a certain time point.When there is a safety accident,it is convenient to locate the position of students in time for rescue.It can also provide position information for analyzing the growth law of students.The system has the advantages of simple operation,stable structure and complete functions.It meets the requirements of campus safety supervision system and can effectively reduce the hidden dangers of campus safety accidents.It has broad application prospects.