Detection of Driver Distraction using YOLOv5 Network

Abstract

The impact of deaths and injuries on families and society is increasing the willingness of the authorities to investigate the cause of the increasing traffic accidents day by day. It is important to emphasize that the majority of traffic accidents are attributed to driver distraction. The role of mobile phones and cigarette usage on driver distraction is a well-known fact. In this study, the authors focus on detecting mobile phone usage and smoking in-vehicle environment. The images collected with a mobile phone docked on the windshield and a yolov5s network is trained with manually labeled images. As a consequence of the study, the authors achieved to distinguish drivers and passengers. Also, they investigate to improve detecting other labeled classes such as ‘DriverHand’, ‘PassengerHand’, ‘DriverHandWithPhone’, etc.