基于嵌入式设备的课堂学生行为检测算法  

Students’Classroom Behavior Detection Algorithm Based on Embedded Devices

作　　者：豆文祥 张峰[1] 张士文[1] 陈政宇 Dou Wenxiang;Zhang Feng;Zhang Shiwen;Chen Zhengyu(School of Electronic Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《电气自动化》2025年第2期101-103,107,共4页Electrical Automation

摘　　要：课堂学生行为分析是教学评估中的重要一环,而传统的行为分析主要通过视频监控或现场实时观察,其效率较低且耗时耗力。为提高算法检测的准确度并满足部署在嵌入式设备上的应用需求,提出了一种改进的YOLOv8目标检测算法。改进的网络结构提高了模型推理速度,并引入注意力机制,对被遮挡行为有更好的识别检测能力。将整个网络部署于嵌入式开发板,通过模型量化和多线程推理加快网络推理速度,完成课堂学生行为的实时检测。试验结果表明,该网络的识别准确率达到89.5%,相较于原始算法提高了1.1个百分点,且推理速度提高约16.2%。因此,所提算法可以在嵌入式设备上实现对学生行为的实时检测。The students’classroom behavior analysis is an important part of teaching evaluation,while the traditional behavior analyses mainly rely on video surveillance or on-site real-time observation,which is inefficient and time-consuming.To improve the accuracy of algorithm detection and meet the requirements of deployment on embedded devices,an improved YOLOv8 object detection algorithm was proposed.The improved network structure upgraded the inference speed of the model and introduced attention mechanism,which had better recognition and detection capabilities for occluded behaviors.The entire network was deployed on an embedded development board,accelerating network inference speed through model quantification and multi-threaded inference,and completing real-time detection of students’classroom behavior.The experimental results show that the recognition accuracy of the network reaches 89.5%,which is 1.1 percentage point higher than the original algorithm,and the inference speed is increased by about 16.2%.Therefore,the proposed algorithm can achieve real-time detection of students’behavior on embedded devices.

关 键 词：嵌入式设备 模型部署 YOLOv8目标检测算法 注意力机制 

分 类 号：TP274.2[自动化与计算机技术—检测技术与自动化装置]