Classroom behavior detection algorithm based on improved YOLOv8
Students'classroom behaviors directly reflect their learning outcomes.Using deep learning methods to detect stu-dent classroom behaviors can more effectively analyze classroom behaviors and improve teaching efficiency.This paper proposes an identification method based on the YOLOv8 model.By adding the BiFormer module,the model's perception of small targets'features is enhanced,improving its behavior detection capabilities in complex environments.The original upsampling module of the model is then replaced with CARAFE to reduce information loss during the upsampling process and improve the model's detec-tion accuracy.Through experiments,our method achieves a mAP@50 of 93%for common classroom behaviors,which is a 1.5 per-centage points improvement over YOLOv8,resulting in more effective identification of classroom behaviors.
万方数据
