摘要
本文研究基于驾驶员脸部信息的疲劳检测,首先选择MTCCN检测驾驶员人脸,在此基础上使用基于AlexNet模型改进的双流网络实现驾驶疲劳检测,该网络利用同时提取当前RGB图像帧的驾驶员静态疲劳特征和经过稠密光流算法Farneback处理过的光流图片帧的驾驶员动态疲劳特征判断驾驶疲劳。驾驶疲劳检测实验结果表明,基于AlexNet改进的双流网络检测准确率为92.87%。
In this paper,based on the pilot fatigue test of facial information,firstly,MTCCN detection drivers face based on AlexNet model is adopted on the basis of using the improved shuangliu network to achieved driver fatigue detection.The network also is utilized to extract the current pilot static fatigue characteristics of RGB image frames and treated through dense optical flow algorithm farneback optical flow image frame and dynamic fatigue characteristics determine the driver fatigue drivingThe experiment results show that AlexNet dual-stream network detection accuracy is 92.87%.
作者
运杰伦
林欣欣
高扬帆
Yun Jielun;Lin Xinxin;Gao Yangfan(College of Information Engineering,Chang'an University,Xi’an 710000,China)
出处
《单片机与嵌入式系统应用》
2019年第12期62-64,共3页
Microcontrollers & Embedded Systems
基金
河南省交通厅重点项目(220024140173)
关键词
驾驶疲劳
卷积神经网络
双流网络
人脸检测
driving fatigue
convolutional neural network
dual-stream network
face detection
