基于CSI的学生课堂行为识别研究  

Research on student classroom behavior recognition

作　　者：郁林 杨彦侃[1] YU Lin;YANG Yankan(School of Digital and Intelligent Industry,Inner Mongolia University of Science&Technology,Baotou 014010,China)

机构地区：[1]内蒙古科技大学数智产业学院,内蒙古包头014010

出　　处：《内蒙古科技大学学报》2024年第4期378-383,共6页Journal of Inner Mongolia University of Science and Technology

基　　金：内蒙古自治区教育厅基金(ZSZX21135)。

摘　　要：针对当前基于视频监控和传感器数据识别学生课堂行为存在的诸多限制性问题,提出一种基于WiFi信道状态信息结合功率谱密度和深度学习的学生课堂行为识别方法。该方法首先采集学生课堂行为信道状态信息(CSI)数据;然后利用Welch法提取CSI功率谱密度(PSD)特征,并转成CSI-PSD光谱图;最后输入构建的CNN-Inception多尺度卷积神经网络模型进行训练和测试。实验表明:该方法无需滤波便可实现高精度的学生课堂行为识别,平均识别率达到99.00%以上。In view of the limitations of current recognition of students’classroom behavior based on video monitoring and sensor data,a method for students’classroom behavior recognition is proposed here,which is based on WiFi channel state information combined with power spectral density and in-depth learning.This method first collects students’channel state information(CSI)data of class behaviors,then uses Welch method to extract CSI power spectral density(PSD)features,and converts them to CSI-PSD spectrogram.Finally,the CNN-Inception multi-scale convolution neural network model is input for training and testing.The experimental results show that this method can achieve high-accuracy on student class behavior recognition without filtering,and the average recognition rate is higher than 99%.

关 键 词：行为识别 信道状态信息 功率谱密度 WELCH法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]