Dangerous Driving Behavior Monitoring System based on Raspberry PI 5

According to the domestic and foreign highway traffic safety accident reports, driver fatigue driving and distracted driving are the main causes of traffic safety accidents, and the concealment and severity of accidents are much higher than others. Therefore, vehicle driving behavior monitoring plays an important role in reducing traffic accidents. This paper proposes a method based on image processing technology to detect fatigue driving and distracted driving. For the former, this paper uses the 68-aspect feature point detection algorithm encapsulated in the Dlib library, and extracts facial fatigue information through the shape detector of Dlib for face recognition and eye and mouth location tracking. For the latter, this paper proposes a driving behavior recognition method based on YOLOv5-Lite, a lightweight object detection model, which still has a good detection effect while reducing the calculation amount, so that it can be easily deployed to edge devices. The Python Flask micro framework and FRP (Fast Reverse Proxy) Intranet penetration tool are used to realize remote real-time web monitoring of drivers, complete the overall function of the system, and can effectively reduce road traffic safety accidents.