AN INTELLIGENT-BASED ALGORITHM FOR DETERMINING DROWSY DRIVERS AND PREVENTION OF ROAD ACCIDENTS

Driving demands full attention and focus, as any decrease in these could lead to accidents. Drowsiness, a condition that reduces a motorist's alertness, poses a significant risk, with delayed detection often resulting in accidents. Traditional methods for detecting drowsiness typically rely on sensors or invasive monitoring, which may not always be practical in everyday situations. This research aims to develop an intelligent algorithm for detecting drowsy drivers and preventing road accidents. The study adopts a physiological approach, utilizing computer vision and deep learning techniques to monitor and identify signs of drowsiness in real-time. The proposed method employs camera-based monitoring to detect symptoms such as yawning, heavy eyelids, and closed eyes from a large dataset of images of drowsy drivers. By using the advanced object detection model YOLOv8s, the system takes advantage of its real-time, high-accuracy facial landmark detection to analyze and assess the drowsiness levels of individuals. The model’s performance, with precision at 0.895, recall at 0.914, and an F1 score of 0.9, demonstrates its effectiveness in accurately identifying drowsiness across various environmental conditions. This non-intrusive, efficient approach provides a viable solution for real-time, low-cost drowsiness detection, offering promising applications in driver’s safety and autonomous vehicle technology.