摘要
学生课堂行为分析可实时反映出学生的课堂学习情况,为优化课堂教学策略和改进教学方法提供了重要依据,探索如何利用机器实时地检测与识别学生课堂行为是智慧教室的重要任务。以视觉显著性的视觉心理学动机为切入点,引入全局视觉显著性机制和BING特征,快速估计课堂视频流中的Objectness学生对象,采用Faster-RCNN模型和时空网络算法对8种典型学生课堂行为进行检测和识别,最后进行全样本学生课堂行为分析。实验结果表明,视觉注意力机制BING特征的引入能大幅度提升算法的实时数据处理能力,计算效率提升近80倍,约为12.5f/s,算法并对8种典型学生课堂行为的检测和识别具有较高的识别精度,平均准确率为87.6%。
The analysis of students’class behavior can reflect students’classroom learning situation in real time,which provides a significant foundation for optimizing classroom teaching strategies and improving teaching methods.It is an important task for smart classroom to explore how to use machines to detect and identify students’class behaviors in real time.This paper is motivated by visual psychological of visual salience and adopt gloal-visual attention mechanism and BING feature to rapidly estimate the class video stream of the Objectness student target.Specifically,we employ Faster-RCNN model and Space-time network algorithm to dectect and identify eight classical class behaviors of student.The experimental results show that the introduction of BING feature of visual attention mechanism can favorably improve the real-time data processing ability of the algorithm,and the computational efficiency is improved nearly 80 times,which is about 12.5 f/s.The algorithm also has a high recognition accuracy for the detection and recognition of eight classical students’class behaviors,with an average accuracy of 87.6%.
作者
夏道勋
田星瑜
唐胜男
XIA Daoxun;TIAN Xingyu;TANG Shengnan(School of Big Data and Computer Science,Guizhou Normal University,Guiyang,Guizhou 550025,China;The Engineering Laboratory for Applied Technology of Big Data in Education in Guizhou Province,Guizhou Normal University,Guiyang,Guizhou 550025,China)
出处
《贵州师范大学学报（自然科学版）》
CAS
2021年第4期83-89,120,共8页
Journal of Guizhou Normal University：Natural Sciences
基金
国家自然科学基金项目(61762023)
贵州省教育科学规划课题(课题编号:2016A055)。
关键词
视觉注意力
学生课堂行为检测
学生课堂行为识别
机器视觉
行为分析
visual attention
detection of students behavior
identification of students behavior
machine vision
behavior analysis
