基于面部特征的听障儿童课堂表现效果预测  

Prediction of Classroom Performance of Hearing-impaired Children Based on Facial Features

作　　者：廖盛斌 谢鹏航 廖庆洲 LIAO Shengbin;XIE Penghang;LIAO Qingzhou(National Engineering Research Center for E-Learning,Central China Normal University,Wuhan 430079;School of Arts and Humanities,Wuhan Vocational College of Software and Engineering,Wuhan 430205)

机构地区：[1]华中师范大学国家数字化学习工程技术研究中心,武汉430079 [2]武汉软件工程职业学院人文学院,武汉430205

出　　处：《计算机与数字工程》2024年第9期2685-2690,2709,共7页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:62077023,61937001);华中师范大学中央高校基本科研业务费项目(编号:CCNU19TS021)资助。

摘　　要：听障儿童由于其听觉有不同程度的受损,他们的课堂教学组织形式和表现与传统课堂有很大的区别。教师对听障儿童在课堂上的行为和表现通常采用手工记录和主观评判。论文在详细分析听障儿童课堂教学组成基础上,通过听障儿童课堂视频数据获取听障儿童在课堂上的对话和变化的面部表情。然后通过多元线性回归将这些特征与课堂表现相关联,提出了使用听障儿童面部特征来预测课堂表现效果的机器学习方法。将听障儿童在课堂时的面部特征与教师的听障儿童课堂表现记录数据相结合,使用多元线性回归来对听障儿童的课堂表现效果进行预测,最终达到了良好的预测效果。Since hearing impaired children suffer from varying degrees of hearing loss,their classroom teaching organization form and performance are very different from the traditional classroom.Teachers usually record the behavior and performance of hearing impaired children in class by hand and make subjective judgments.The composition of classroom teaching for hearing-impaired children is first presented in this paper,the conversation and changing facial expressions of the hearing impaired children are obtained through the video data in class.And then the multiple linear regression is used to correlate these characteristics with classroom performance,a machine learning method using facial features of hearing-impaired children to predict classroom performance is proposed.The method achieves good prediction results by combining the facial features of hearing impaired children with the data recorded by teachers in class in multiple linear regression model.

关 键 词：听障儿童 预测 课堂表现 线性回归 

分 类 号：G40-057[文化科学—教育学原理]