Design And Implementation Of Laboratory Attendance Management System Based On Web And Face Recognition Technology

As the front position of scientific research development,university laboratory construction is bound to develop in the direction of informatization and intelligence under the background of information age.Efficient attendance management is an important guarantee for the quality of laboratory teaching activities.This paper takes the undergraduate course laboratory as the research object,through the investigation of the laboratory situation,summarizes the main problems existing in the current laboratory attendance mode:1.Undergraduate course laboratory has always been a weak link in the construction of colleges and universities,and there is no perfect attendance management system.Most of the undergraduate course laboratory construction is lack of scientific research funding support.If we purchase the mainstream attendance equipment in the market in large quantities,it will undoubtedly increase the research funding burden.2.At present,most of the undergraduate experimental courses are signed in by traditional signature or roll call,which leads to dishonest behaviors such as signing on behalf of others.3.In the case of a large number of experimental courses,if the roll call or check-in is carried out by means of full coverage,the check-in link may cost too much time.In view of the above problems,this paper decides to combine face recognition technology with web technology to realize a B/S mode attendance management system.Before the system development,the detailed requirements analysis is carried out,including the application scenario analysis,system function analysis and system performance analysis.Through the demand analysis,the system is divided into two parts:one is face recognition module and the other is management system.Firstly,the face recognition module realizes a fast face detection and key point detection network without anchor.The model size of the method is only 7.3M.Then,the five key points extracted from face detection algorithm are aligned simply and quickly by affine transformation.Then,through introducing dense connection to MobileNet v2 network structure,and optimizing loss function and target function,a lightweight face recognition algorithm is realized.The reasoning speed of i7-4790 can reach 50ms,and the recognition accuracy of LFW data set is 99.13%.The design of face recognition module is completed by combining these combinations.This paper designs face recognition module without relying on GPU equipment,which can greatly reduce the cost of system development,and has strong robustness to occlusion problems,small faces,multiple faces,etc.Through the requirement analysis,the management system adopts B/S design mode,the MVC framework adopted in the system framework,and designs the database in detail.Finally,the face recognition module is embedded in the management system,and the design and implementation of the complete attendance management system are realized.Because the system of this paper adopts B/S design mode,it is unnecessary to access the system without specific client software.In the topology design of the system,the back-end server and algorithm server are deployed separately.Such a design can reduce the difficulty of upgrading and maintenance in the later stage of the system,and improve the robustness,maintainability and scalability of the system.