A Comparative Study of Four YOLO-Based Models for Distracted Driving Detection

In general, the accidents are often attributed to the decline of driving skills and reduced attention among the aging population. Moreover, the rise of infotainment features in vehicles has led to loss of concentration while driving. To address this issue, we previously developed an intelligent driver assistance system to identify instances of distracted driving. This paper focuses on enhancing the detection accuracy of distracted driving by utilizing an upgraded version of our system equipped with different detectors. The system is designed to recognize various distracted driving behaviors, particularly focusing on the driver hand movements. From the evaluation, we observed that while models like YOLOv8 show improvements in detecting specific actions like steering wheel and IVI operations, challenges remain in accurately identifying other crucial activities such as cell phone usage and general hand movements.