Accident Avoidance System using Drowsiness and Pothole Detection

Road accidents due to driver drowsiness and poor road conditions, such as potholes, remain significant safety concerns. This paper presents an AI-powered accident avoidance system that integrates real-time driver drowsiness detection and pothole identification to enhance road safety. The system employs Dlib's facial landmark detection and the Eye Aspect Ratio (EAR) metric to assess driver fatigue, triggering alerts when drowsiness is detected. For pothole detection, the YOLO (You Only Look Once) deep learning model is utilized, trained on a custom dataset to identify road hazards even under low-light conditions. The integration of these two models provides a comprehensive solution capable of real-time monitoring, ensuring both driver awareness and road condition assessment. Performance evaluation demonstrates high accuracy in both drowsiness detection and pothole identification, proving the system’s effectiveness in mitigating road-related accidents. This research contributes to the development of smart transportation solutions aimed at reducing accidents and enhancing driver safety.