基于视频分析的学生网课疲劳检测方法

Fatigue detection method of students’online class based on video analysis

为了更好地提高学生在远程教育中的学习效率,避免学生出现走神瞌睡等情况,文中提出了一种基于视频分析的学生眼睛疲劳状态的检测方法。该方法的原理是将疲劳检测视为基于图像的序列识别。首先采用深度级联的多任务框架从视频中提取人眼区域;然后通过深层卷积学习空间特征,并通过长-短期记忆单元分析相邻帧之间的关系;最后,对学生状态进行序列级预测。实验结果表明,与现有的方法相比,所提出的方法具有较高的准确率和精度。

In order to better improve the learning efficiency of students in distance education,and to prevent students from getting drowsy and so on,a method for detecting fatigue states based on the eye states of students is proposed here.The principle of this method is to treat fatigue detection as image-based sequence recognition.First,a deep cascading multi-task framework is used to extract the human eye area from the video.Then it learns spatial features through deep convolution,and analyzes the relationship between adjacent frames through long-short-term memory units.Finally,a sequence-level prediction is made on the state of the students.Experiment results show that compared with the existing methods,the proposed method has higher accuracy and precision.