基于计算机视觉的课堂专注度检测研究

Engagement Detection for Offline Education Based on Computer Vision

为了检测课堂学习时班级整体专注程度,设计了一套基于计算机视觉的课堂专注度检测系统。该系统根据面部关键点位置计算头部偏转角度,进而判断视线是否位于专注区域,采用ViViT模型对学习者的专注度进行四分类,最后综合人脸检测信息、视线信息、专注度分类信息计算班级整体专注度。实验结果证明,该系统能根据监控视频快速、有效地评估班级在某时刻的整体专注度,为教师提供客观、准确的专注度数据,有助于教师有针对性地及时调整教学方式,提高教学质量。

In order to detect the overall engagement level of the class during classroom learning,a computer vision-based classroom engagement detection system is designed.The head deflection angle is calculated based on the location of key points on the face,which is used to determine whether the line of sight is located in the engagement area.The ViViT model is adopted to perform four classifications of learners′engagement.Face detection information,line of sight information,and engagement classification information are finally combined to calculate the overall engagement of the class.The experimental results prove that the system can quickly and effectively evaluate the overall engagement of the class at a certain moment based on the monitoring video.It can provide teachers with objective and accurate engagement data,which can help teachers adjust teaching methods in a targeted and timely manner to improve teaching quality.