Enhanced Driver Drowsiness Detection Using Deep Multi-Scale Convolutional Neural Network

Driver fatigue is a significant contributor to vehicular accidents, highlighting the necessity for precise, real-time monitoring technologies to enhance road safety. This research introduces an innovative driver drowsiness detection technique using a Deep Multiscale Convolutional Neural Network (DMS- CNN), aimed at accurately recognizing subtle indicators of fatigue, including facial expressions, eye closure, and head motions. The DMS-CNN utilizes multiscale feature extraction to capture subtle variations in diverse lighting and driving situations. The model achieved a precision of 96%, a precision of 95%, and a recall of 94% using a diversified dataset, markedly surpassing conventional single-scale CNNs in managing variable- scale features and environmental obstacles. The experimental findings confirm the robustness and dependability of the DMS- CNN, establishing it as an excellent option for real-time driver monitoring.