Eye Tracking based Driver Fatigue Monitoring System

Abstract

Drowsiness and Fatigue of drivers are significant causes of road accidents. Every year these increase death rate globally. Therefore, we have proposed a system called Eyes tracking based driver fatigue monitoring system that takes visual information and uses artificial intelligence and computer vision to detect driver’s drowsiness states. System will help us in reducing accidents around the global. We aim to use computer vision algorithms to track, locate and analyse the driver’s face and eyes movement to measure PERCLOS, a scientifically supported measure of drowsiness associated with slow eye closure. System will detect user’s face and region of interest (EYES) by using CV algorithms and then feed image to Neutral network to get image classified to alert the driver. Based, on classified image system will detects driver’s drowsiness if he is in a drowsiness state or not. This system works well in day light and in night too if infrared cameras are used. For making, the system we have trained CNN architecture to get image classified and to detect features we have used CV algorithms and Dlib library. For more accurate we marge both CNN and Dlib for detection.