深度学习驱动的课堂行为自动识别技术与方法综述

打开文本图片集

中图分类号：TP391.4 文献标识码：A 文章编号：2096-4706（2025）20-0109-06

Abstract： Deep learning is graduall replacing traditional Machine Learning methods as thecore tool forautomatically identifyingclassrohavis.Fistlyispperdeiesstudets'posiiveadnegatiebevirs，andelaboratesoeeal processofautomaticallydentifingclasrooeviors.econdlyitefltroestetheclassictwoksineldof Deep Learing such as Convolutional Neural Networks （CNN），Recurrent Neural Networks （RNN），and Graph Convolutional Networks （GCN）.Itfocuses on exploringthecurrent aplication status of two keytechnologies，namelyobject detectionand poserecognitioniefeldofasroomhavioecogtioninalymmaresthatagesandlitatiosoeep Learning in automatic clasroom behavior recognition，and proposes future research directions basedon curent research progress，aiming to provide reference for scholars and researchers in related fields.

Keywords： Deep Learming; classroom behavior recognition; Object Detection; posture recognition

0 引言

随着数字技术在课堂教学质量评价领域的广泛应用，传统的课堂教学质量评价方式逐渐显现其局限性，无法满足数字时代课堂教学质量评价的发展需求。（剩余9674字）