Analysis of Students’ Class Status Based on Deep Learning

Aiming at the problem that it is difficult for teachers to obtain accurate classroom status in daily teaching process, which is not conducive to making targeted adjustments to teaching methods, this paper proposes a deep learning-based student class status analysis system. The system uses camera to capture classroom video, and uses image recognition, target detection, deep learning and other technologies to detect the behavioral state of students in the classroom in real time and concentration information, and through the statistical analysis of the collected data, it helps the teacher to get timely feedback in the classroom, and better judge the learning state and concentration of students. In order to realize the real-time and accuracy of the system design, the system in the paper introduces the OpenPose model into the YOLOv5 network to identify the students’ skeletal keypoints, and synthesizes the results of the processing of the YOLOv5 model and the OpenPose model to make an analysis of the students’ classroom behaviors and concentration. The experimental results show that the loss curve can achieve good convergence, and the AP and mAP can reach 95.1% and 88.0%, respectively.