• 首页
  • 期刊简介
  • 编委会
  • 投稿指南
  • 收录情况
  • 杂志订阅
  • 联系我们
引用本文:王禹钧,马致明.基于深度学习的学生课堂行为识别研究[J].软件工程,2023,(7):40-43.【点击复制】
【打印本页】   【下载PDF全文】   【查看/发表评论】  【下载PDF阅读器】  
←前一篇|后一篇→ 过刊浏览
分享到: 微信 更多
基于深度学习的学生课堂行为识别研究
王禹钧, 马致明
(新疆师范大学计算机科学技术学院, 新疆 乌鲁木齐 830054)
627885569@qq.com; 406287175@qq.com
摘 要: 智慧课堂面临学校硬件设备难以普适、课堂数据缺失等问题,给学生课堂行为识别研究带来了一定的难度和挑战。为了降低对课堂场景部署设备的要求,提高识别精度与速度,在已有识别模型基础上提出了一种基于深度学习的学生课堂行为识别模型。首先融合重影模块(Ghost)实现轻量化,其次加入坐标注意力机制(Coordinate Attention)提升检测精度。实验结果表明,改进后的模型识别精度(mAP)达到了86.2%,与原模型相比提高了3.5%,推理时间减少了16.7%,参数量降低了35.5%,速度与精度均有一定提升,符合智慧课堂的基本要求。
关键词: 深度学习;学生课堂行为;行为识别;智慧课堂
中图分类号: TP3-05    文献标识码: A
基金项目: 新疆维吾尔自治区自然科学基金项目“基于认识学习的多模态问题语义分析与研究”(2022D01A227).
Research on Classroom Behavior Recognition of Students Based on Deep Learning
WANG Yujun, MA Zhiming
(School of Computer Science and Technology, Xinjiang Normal University, Urumqi 830054, China)
627885569@qq.com; 406287175@qq.com
Abstract: Smart classrooms face problems such as difficulty in adapting school hardware equipment and missing classroom data, which poses certain difficulties and challenges to the research of student classroom behavior recognition. In order to reduce the requirements for deploying equipment in classroom scenarios and improve recognition accuracy and speed, this paper proposes a deep learning-based student classroom behavior recognition model based on existing recognition models. Firstly, the Ghost module is integrated to achieve lightweight, and secondly, a Coordinate Attention mechanism is added to improve detection accuracy. The experimental results show that the improved model has a recognition accuracy (mAP) of 86.2% , which is 3.5% higher than the original model. The inference time is reduced by 16.7% , and the parameter quantity is reduced by 35.5% . The speed and accuracy are both improved to a certain extent, which meets the basic requirements of smart classrooms.
Keywords: deep learning; student classroom behavior; behavior recognition; smart classroom


版权所有:软件工程杂志社
地址:辽宁省沈阳市浑南区新秀街2号 邮政编码:110179
电话:0411-84767887 传真:0411-84835089 Email:semagazine@neusoft.edu.cn
备案号:辽ICP备17007376号-1
技术支持:北京勤云科技发展有限公司

用微信扫一扫

用微信扫一扫