基于深度学习的驾驶员疲劳驾驶检测研究  

Research on Driver Fatigue Detection Based on Deep Learning

作　　者：张振利 李亮 宋京京 付豪 陈源 Zhang Zhenli;Li Liang;Song Jingjing;Fu Hao;Chen Yuan(School of Electrical Engineering and Automation,Jiangxi University of Science and Technology,Ganzhou,Jiangxi 341000,China)

机构地区：[1]江西理工大学电气工程与自动化学院,江西赣州341000

出　　处：《机电工程技术》2024年第1期193-197,共5页Mechanical & Electrical Engineering Technology

基　　金：江西省创新基金资助项目(YC2021-S588)。

摘　　要：为解决驾驶过程中出现疲劳的问题,提出了一种基于卷积神经网络的疲劳驾驶检测算法。在原始YOLOv5网络的基础上,提出一种Mosaic-8数据增强的方法提高网络训练的学习效率;并在骨干网络C3层中引入SE注意力机制模块,增加网络中通道之间的相关性;最后通过EIOU损失函数替换原网络中的损失函数。相比于原始算法,改进算法对人体的眼睛和嘴巴部位有着更好的检测效果,可以准确地提取驾驶员的眼睛嘴巴开合特征。在驾驶疲劳的评判标准上选择PERCLOS指标对驾驶员的疲劳进行判定,能够客观准确地对驾驶员的疲劳状态进行预警。相较于其他深度学习的疲劳检测算法,所提方法在精度上有所提高,实验证明其准确率可以达到97.8%。In order to solve the problem of fatigue in the process of driving,this paper proposes an algorithm based on convolutional neural network for fatigue driving detection.Based on the original YOLOv5 network,a Mosaic-8 data enhancement method is proposed to improve the learning efficiency of network training.The SE attention mechanism module is introduced into C3 layer of backbone network to increase the correlation between channels in the network.Finally,the EIOU loss function is used to replace the original network loss function.Compared with the original algorithm,the improved algorithm has a better detection effect on the human eyes and mouth,and can accurately extract the opening and closing features of the driver's eyes and mouth.In the evaluation criteria of driver fatigue,PERCLOS index is selected to judge driver fatigue,which can objectively and accurately give early warning to the fatigue state of drivers.Compared with other deep learning fatigue detection algorithms,the accuracy of the proposed method is improved.Experiments show that the accuracy of the proposed method can reach 97.8%.

关 键 词：疲劳驾驶检测 YOLOv5 深度学习 PERCLOS 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程] U463.6[机械工程—车辆工程]