Research And Development Of The Smart Classroom Video Analysis System Based On Object Detection

The modernization and informationization of education is a crucial step in the development of China’s education,and the construction and use of smart classroom is the foundation of education informationization.With the development of emerging technologies,cloud computing,big data,artificial intelligence and other technologies are advancing hand in hand,which supports the deep integration of intelligent classroom,an information technology means,and talent training system.Like other key universities in China,Shandong University has built most of its classrooms into smart classrooms,enabling teachers to broadcast and record lectures in smart classrooms.However,the existing cloud platform of smart classroom is not perfect in the application of hardware devices and course video data,and the relevant software application is not rich and intelligent enough.A large number of course videos are stored on the Smart Classroom cloud platform.In addition to facilitating students’ self-study,the school is also looking for ways to make it provide big data support for teaching management,students’ course selection and teaching evaluation.In view of the problem that the existing smart classroom cloud platform application software cannot meet various teaching management requirements,this paper designs and develops a smart classroom video analysis system based on the big data set of course video of our smart classroom cloud platform and for teaching management requirements.The system uses the object detection technology to process the course video images,extract the data information in the images,and present it to the system users in a concise and objective way.As an auxiliary means,it helps teaching administrators to make more objective and accurate teaching evaluation.The main research work of this paper includes:(1)Elaborated the research,design and implementation process of the video analysis system of smart classroom.Through literature research and data processing of smart classroom at home and abroad,the development process of smart classroom is deeply understood.Then,after full discussion with the construction and management personnel of smart classroom,demand analysis and overall architecture design of the system are carried out,and five functional modules of the system are determined:teacher management module,course management module,classroom management module,user management module and video processing module.The functional modules of the system are designed and implemented in detail and the key technologies involved are introduced.(2)For the video processing module,the object detection technology based on deep learning is adopted for image processing of the smart classroom video.A separate data set is made for the multichannel video data and application scenarios of the smart classroom,and the BiFPN module is integrated into the YOLOv5s model to form the BiFPN-Yolov5s network model that is more suitable for the scenarios in this paper,which increases the accuracy rate by 2.1%compared with the original model without decreasing the speed.(3)The system adopts B/S architecture and adopts the way of front and back end separation for development.The React framework,Ant Design and other technologies are used for front-end development,and the Spring Boot and MyBatis frameworks are used for back-end development.In order to give full play to the advantages of GPU image processing,the video processing module adopts the Python language Flask framework and the BiFPN-YOLOv5 network model studied in this paper is deployed on the GPU server,and inter-system communication is conducted through the RESTFul interface.Finally,the system is tested by unit test and black box test to ensure the system’s complete function and smooth operation.The video analysis system of smart classroom constructed in this paper,as an auxiliary means of data processing of smart classroom,can provide more comprehensive and accurate classroom information for teachers and students,provide objective and comprehensive classroom data for teaching administrators in teaching evaluation,and help students to know about the classroom status of a certain course and a certain teacher in the early stage of course selection,and determine whether to choose a course.The research work of this paper is a meaningful attempt to analyze the video data of intelligent classroom.