Design And Implementation Of Online Project Review System Based On WebRTC And Face Recognition

With the continuous advancement of social informatization construction,in order to solve the problems of long time and low efficiency in project review,and the inability to defend offline in the era of coexistence of epidemics,this thesis designs and implements an online project review system based on Web RTC and face recognition.This system can complete all the links in the project review process without installing any client or plug-in,and realize functions including face recognition login verification,web-side video conference and other functions.The main research and work of this thesis are as follows:(1)Research on the web-side video conferencing technology,and design and implement a Web RTC-based real-time audio and video communication framework.Firstly,the communication architecture of the video conference is designed,the SFU architecture is adopted and a Kurento media server is built to realize the processing and forwarding of audio and video streams;a session management mechanism is designed for common conference signaling,and session management,media Negotiation and network negotiation;studied the process of establishing connection for audio and video real-time communication based on Web RTC in SFU mode,and designed and realized the instant message communication based on conference room.Finally,the performance of the audio and video real-time communication based on Web RTC is analyzed,and the results show that the audio and video communication realized by this system can meet the needs of the web-side video conference.(2)Aiming at the identity authentication problem of online review,a face verification optimization method based on evolutionary convolutional neural network is proposed.Firstly,the adaptive strategy of the genetic algorithm is improved.For the problem that the genetic operator is easy to fail when the convolutional neural network is mapped to the chromosome,a new coding method is adopted,and the improved genetic algorithm and the new coding method are used.To optimize the parameters of the convolutional neural network globally,and find the best network parameters through continuous evolution.The experimental results show that the proposed evolutionary convolutional neural network can obtain better face recognition accuracy,and the improved genetic algorithm can effectively improve the speed of finding the global optimal solution.(3)According to the relevant requirements of the project review system,the role use cases and business processes are determined,and the system architecture and database entities are designed and implemented.The system is divided into face verification module,project review module and video conferencing module according to functions,a lightweight convolutional neural network is designed and a self-built data set is used for training,and the face recognition login verification based on the web side is realized;using Java Web technology The business logic of project declaration and review is realized;the real-time audio and video communication and instant message communication based on the Web RTC framework are realized for the online defense function.Finally,the compatibility of the system is analyzed,test cases are written and the system is fully functionally tested.The results show that under the condition of normal network bandwidth,the functions of the system are normal,and it has good cross-platform,which can meet the requirements of use.