Improved YOLOv5 for skeleton-based classroom behavior recognition

Abstract

Classroom behavior is an important criterion for evaluating instructional efficacy. In comparison to other behaviors, the challenge of classroom behavior detection is primarily influenced by ambient light variables and the presence of too many targets to recognize, resulting in missed detection. Recent research has demonstrated that information about the human skeleton can be used to identify classroom conduct. As a result, we present an enhanced yolov5-based skeletal recognition system for detecting classroom behavior in this paper. First, the YOLOv5 detection algorithm is improved to extract target prospects for the problem of missed detection; then, the human skeleton information is obtained using the Alphapose framework; finally, the skeletal data is sent into a two-stream adaptive graph convolution network to allow for the accurate recognition of various classroom behaviors. According to extensive tests, the detection algorithm based on bone recognition improves detection accuracy and lowers the false detection rate.

An Improved YOLOv7-CO Classroom Behavior Detection Algorithm Combining Coordinate Attention Mechanism and Omni-Dimensional Dynamic Convolution

“…Traditional human observation methods are not capable of quickly and accurately capturing a large number of student classroom behaviors. However, numerous studies and practices have shown that using deep learning models for classroom behavior analysis is an effective way [1][2][3][4]. At the same time, some scholars have provided great research ideas for behavior detection by constructing virtual scenes through Digital Twin technology to obtain relevant information [5][6].…”

Students Behavior Detection Based on Improved YOLOv5 Algorithm Combining with CBAM Attention Mechanism

“…However, a drawback of their method was its reliance on addressing the partial blockage of numerous human targets, and the inclusion of the skeletal (2S-AGCN) module resulted in slower inference and detection times. In our proposed approach, we aimed to strike a balance between model speed and accuracy by modifying all parts of the model [12].…”