Detection of Driver Drowsiness Using Adaptive Eye Characteristic Ratio for Enhanced Road Safety.

Driver drowsiness is one of the major causes of road accidents worldwide and poses a serious threat to public safety. Fatigue-induced impairment affects a driver’s alertness, reaction time, and decision-making ability, significantly increasing the probability of accidents. This paper presents a real-time, non-intrusive driver drowsiness detection system based on an Adaptive Eye Characteristic Ratio (AECR) combined with a Vision Transformer (ViT) model. A standard webcam is used to continuously monitor the driver’s facial and eye movements. The proposed approach focuses on detecting prolonged eye closure, which is a strong indicator of drowsiness. The Vision Transformer model is trained using the backpropagation algorithm with AdamW optimization to improve accuracy and regularization. A secure web-based application developed using Flask displays bounding boxes around the driver’s eye region when drowsiness is detected. Experimental results demonstrate that the proposed system achieves high accuracy and robustness under varying lighting conditions. The system is cost-effective, easy to deploy, and suitable for real-world vehicular applications, making it a promising solution for enhancing road safety and preventing fatigue-related accidents.