基于深度学习的学习者在线课堂参与度研究  

Research on Learner’s Online Classroom Engagement Degree Based on Deep Learning

作　　者：齐永锋 周晨阳 QI Yong-feng;ZHOU Chen-yang(College of Computer Science and Engineering,Northwest Normal University,Lanzhou,Gansu,China 730070)

机构地区：[1]西北师范大学计算机科学与工程学院,甘肃兰州730070

出　　处：《现代教育技术》2021年第12期42-50,共9页Modern Educational Technology

摘　　要：文章针对在线教育环境中学习者监管缺失、教学反馈滞后等问题,构建了包含课后实时答题、头部姿态估计、表情识别的三维课堂参与度信息融合框架。为验证此框架的有效性,文章通过比较框架对学习者给出的课堂参与度评分与其填写的NSSE-China问卷评分求取绝对误差之和,并进行参与度前提与两者间绝对误差之和的多次对比实验,结果表明:参与度前提与两者间绝对误差之和存在强烈的负相关性,说明此框架有效。研究基于深度学习的学习者在线课堂参与度,有助于提高在线教育的课堂教学质量,并助力现代化信息教育的发展。Aiming at the issues of lacking learner supervision,and lagging teaching feedback in the online education environment,this paper constructed a three-dimensional information fusion framework of classroom engagement degree including real-time after-class answering,head pose estimation,and expression recognition.In order to verify the validity of this framework,the paper calculated the sum of absolute errors by comparing the classroom engagement degree scores given by the framework to learners and scores of the NSSE-China questionnaire they filled out,and conducted several comparative experiments for the prerequisite of engagement and the sum of absolute errors between the two.The results showed that there was a strong negative correlation between the prerequisite of engagement degree and the sum of absolute errors between the two,indicating the effectiveness of the framework.Researching on learners’online classroom engagement degree based on deep learning could help improve the teaching quality of online education and contribute to the development of modern information education.

关 键 词：在线教育 课堂参与度 头部姿态估计 人脸表情识别 深度学习 

分 类 号：G40-057[文化科学—教育学原理]