基于深度神经网络的学生课堂行为识别研究  

Students′classroom behavior recognition based on deep neural network

作　　者：朱霞 李明星 章翔飞[3] ZHU Xia;LI Mingxing;ZHANG Xiangfei(School of Computer,Jiangsu University of Science and Technology,Zhenjiang 212100,China;Jingjiang College,Jiangsu University,Zhenjiang 212013,China;Information Management Office,Jiangsu University of Science and Technology,Zhenjiang 212100,China)

机构地区：[1]江苏科技大学计算机学院,镇江212100 [2]江苏大学京江学院,镇江212013 [3]江苏科技大学信息化管理办公室,镇江212100

出　　处：《江苏科技大学学报（自然科学版）》2023年第6期72-76,共5页Journal of Jiangsu University of Science and Technology:Natural Science Edition

基　　金：国家自然科学基金项目资助(62276118,61772244)。

摘　　要：在学生课堂行为识别场景下,存在干扰因素较多且缺少公开的数据集,识别效果较差问题,提出一种融合行为姿态和注意力机制的高校学生课堂行为识别模型,利用注意力机制的残差卷积神经网络,提取教学视频中学生课堂行为的空间和通道特征,利用与视觉描述互补的行为姿态信息获取深度特征蕴含的高层行为信息.为验证模型的有效性,构建了学生课堂行为数据集进行实验,实验结果表明,所提模型在高校学生课堂行为识别中展现了较好的性能.Current mainstream behavior recognition methods prove ineffective in the challenging context of recognizing students′classroom behavior due to scene complexity,numerous interfering factors,and lack of publicdatasets.To address these limitations,this paper presents a novel recognition model that integrates behavioral poseinformation and attention mechanisms.The proposed model employs a residual convolutional neural network with an attention mechanism to extract spatial and channel features from teaching videos,while leveraging behavioral pose information to enhance the description of visual cues and capture high-level behavioral information embedded in depth features.To evaluate the effectiveness of the model,we construct a comprehensive dataset of students′classroom behavior and conduct extensive experiments.The results demonstrate that the proposed model achieves high performance in college students′classroom behavior recognition.

关 键 词：深度学习 注意力机制 卷积神经网络 课堂行为识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]