摘要
目前疲劳检测主要是通过眼部PERCLOS值来判断,但是这种方法检测疲劳特征较为单一,影响了检测的准确率。本文提出一种基于卷积神经网络的多特征融合学生疲劳检测算法,首先用MTCNN对人脸进行关键点定位,在此基础上用人脸归一化的方式精准提取眼睛和嘴部的特征图像并进行眼部定位;其次,构建眼、嘴数据集,完成眼、嘴部状态分类模型训练;最后,用训练好的模型将眼、嘴部的疲劳特征相融合并根据改良的MAR值判断方法进行疲劳检测。实验结果表明该方法的准确率达到了96.2%,实时性也得到了极大的改善。
The fatigue detection is mainly judged by the PERCLOS value of the eye,but this kind of characteristics is relatively single,which affects the accuracy of the detection. In this paper,a multi feature fusion student fatigue detection algorithm based on convolutional neural network is proposed. Firstly,MTCNN is used to locate the key points of the face. On this basis,a face normalization method is proposed to accurately extract the feature images of the eyes and mouth and locate the eyes. Then the eye and mouth dataset is constructed to complete the training of eye and mouth state classification model. Finally,the trained model is used to integrate the fatigue characteristics of eyes and mouth,and the fatigue detection is carried out according to the improved MAR value judgment method. The experimental results show that the accuracy of this method is96.2%,and the real-time performance is greatly improved.
作者
陈藩
施一萍
胡佳玲
谢思雅
刘瑾
CHEN Fan;SHI Yiping;HU Jialing;XIE Siya;LIU Jin(School of Electronic and Electrical engineering,Shanghai University of Engineering and Technology,Shanghai 201620,China)
出处
《智能计算机与应用》
2021年第9期94-98,共5页
Intelligent Computer and Applications
基金
国家自然科学基金(61701296)。