Design and Implementation of Classroom Student Behavior Detection Algorithm Based on Improved YOLOv5s
Aiming at the problems of missed detection and false detection that occur in complex classroom environments and dense student targets,this paper proposes a student classroom behavior detection method that improves the YOLOv5 target detection algorithm.Firstly,a student behavior data set containing six behaviors is created;secondly,a multi-head self-attention mechanism is added to the Backbone and Neck parts to enhance the model's ability to understand complex and crowded spatial relationships in classroom images;finally,the detection head is used to calculate the border of the lost CIoU is replaced by a more comprehensive SIoU.Experimental results show that compared with the original model,the average precision of the improved model is increased by 3.9％,and the recall rate is increased by 2.6％,which improves the model's ability to accurately detect targets.
student behaviormulti-head self-attention mechanismSIoU loss function
万方数据
维普
