安徽工程大学学报2025，Vol.40Issue(2)：39-47,9.
基于改进YOLOv5的课堂教学评估算法与应用
Evaluation Algorithm and Application of Student Classroom Based on Improved YOLOv5
摘要
Abstract
To assist teachers in improving teaching quality,a teaching assistance system that can accurately identify student classroom conditions and conduct reasonable evaluations was constructed.A dual granularity annotated Student Action(SA)dataset was set up,which adopts two classification methods;The paper proposed the SGE-BI(Spatial Group wise Enhancement-BiFormer)dual attention mechanism module and added it to the YOLOv5 algorithm;An improvement on the traditional Analytic Hierarchy Process(AHP)is obtained by combining the student behavior dataset with dual granularity classification,and the Analytic Hierarchy Process Double Granularity Weighted(AHP-DG)is bronght forward.Simulation experiments were conducted on the SA dataset,using the improved YOLOv5 algorithm to improve the accuracy,mAP@0.5 and mAP@0.5:0.95 value increased by 1.9%,0.5%,and 0.8%respectively;Data analysis was conducted on the recognition results obtained from the constructed system on 10 real student classroom videos collected.The AHP-DG method is closer to the evaluation results of experts in the field than the traditional AHP method.Compared with the original algorithm,the improved YOLOv5 algorithm achieves more accurate results in identifying students'classroom behaviors;Compared with the traditional AHP method,the proposed AHP-DG method produced better results in evaluating students'classroom conditions.关键词
课堂状况评估/目标检测/深度学习/卷积神经网络/注意力机制/层次分析法Key words
classroom condition evaluation/object detection/deep learning/convolutional neural network/attention mechanism/analytic hierarchy process分类
计算机与自动化引用本文复制引用
王岱嵘,陈新泉,丁达之..基于改进YOLOv5的课堂教学评估算法与应用[J].安徽工程大学学报,2025,40(2):39-47,9.基金项目
国家自然科学基金面上项目(61976005) （61976005）
安徽省自然科学基金项目(2108085MF213) （2108085MF213）
安徽工程大学校企合作项目(2023qyhz15) （2023qyhz15）
国家级大学生创新训练项目(202210363094) （202210363094）
