基于头部姿态的学习注意力判别研究

A Head-posture Based Learning Attention Assessment Algorithm

学习注意力是学生学习效率的一个重要因素,直接影响学生的学习效果。为了有效监测学生在传统课堂教学中的注意力情况,提出一种基于头部姿态识别的学生注意力判别方法。首先运用卷积神经网络对视频图像的面部特征点进行检测,其次采用比例正交投影迭代变换(pose from orthography and scaling with iterations,POSIT)算法对人脸进行跟踪识别,并对人脸的旋转角度进行计算,根据其头部的倾斜角度对学生注意力进行分析研究。测试结果表明,提出的模型注意力检测准确率为88.7%,可以有效地对学生注意力进行检测,具有较好的应用前景。

Learning attention is an important factor for students’learning efficiency,which directly affects students’learning effects.To effectively monitor students’attention in traditional classroom teaching,a method of judging students’attention based on head gesture recognition was proposed.Firstly,a convolution neural network was used to detect facial feature points on video images.Secondly,the pose from orthography and scaling with iterations(POSIT)algorithm was adopted to track and recognize human faces.At the same time,rotation angles of human faces were computed.The tilt angles of heads were applied to study students’attention.Experimental results show that attention detection accuracy of the proposed model is 88.7%,which can effectively detect students’attention with a good application prospect.