College of Computer Science,Yangtze University,Jingzhou,Hubei 434023 ,China 在期刊界中查找 在百度中查找 在本站中查找
Affiliation:
College of Computer Science,Yangtze University,Jingzhou,Hubei 434023 ,China
Fund Project:
摘要
|
图/表
|
访问统计
|
参考文献
|
相似文献
|
引证文献
|
资源附件
|
文章评论
摘要:
随着科技的不断进步，深度学习算法在学生课堂行为识别领域的重要性日益凸显，吸引了众多学者的广泛关注。然而，受限于传统教学课堂管理手段以及深度学习在教学领域应用的局限性，深度学习在教育中的发展仍面临挑战。基于深度学习的学生课堂行为识别研究，文章以YOLOv8为基础，融合CBAM 注意力机制，结合通道和空间注意力策略，提升了模型的检测能力。实验结果表明，改进后的YOLOv8 CBAM模型相较于原始模型在精度上提高了1%，其mAP（mean Average Precision）达到了92%，并且对学生低头写字、低头看书、抬头听课、站立、教师指导和小组讨论等行为的识别率均达到了90%。
With the continuous advancement of technology， the importance of deep learning algorithms in the field of student classroom behavior recognition has become increasingly prominent， attracting widespread attention from numerous scholars. However， due to the limitations of traditional classroom management methods and the application of deep learning in the teaching field， the development of deep learning in education still faces challenges. The research on student classroom behavior recognition based on deep learning， using YOLOv8 as the foundation， integrates CBAM attention mechanism， and combines channel and spatial attention strategies to enhance the detection ability of the model. The experimental results showed that the improved YOLOv8-CBAM model improved accuracy by 1% compared to the original model， with a mean average precision （mAP） of 92%， and achieved a recognition rate of 90% for behaviors such as students bowing their heads to write， reading， listening， standing， teacher guidance， and group discussions.
