Design and Development of a Fatigue Driving Detection System

A major cause of traffic accidents is fatigue driving, which poses a serious threat to road traffic safety. The design implements a fatigue driving detection system based on deep learning, focusing on research into fatigue driving detection technology. The system employs an improved MobilenetV3-Yolov5 model for facial detection, significantly enhancing detection speed while maintaining high accuracy, making it suitable for real-time in-vehicle environments.By using the DLIB facial landmark detection algorithm, the system can accurately extract key fatigue features such as the driver's eyes, mouth, and head pose angles. Eye features are evaluated using the PERCLOS parameter and blink frequency, mouth features are detected by the MAR parameter for yawning behavior, and head pose angles are assessed for abnormal movements through changes in Euler Angles. Experimental results indicate that the system operates stably under different lighting conditions and complex scenarios, achieving a fatigue detection accuracy rate exceeding 90% and good real-time performance, with detection speeds reaching up to 80 frames per second. Additionally, the system includes functionalities such as data storage, real-time monitoring, and web-based interaction, enabling drivers and administrators to conveniently view and manage detection results in real time. The research in this paper provides effective technical support for the practical application of fatigue driving detection technology and lays the groundwork for subsequent system optimization and deployment on cloud platforms, as well as smart transportation systems.