Driver Eye Tracking Over Iot

Abstract

Naturalistic driving studies have shown that a driver’s allocation of visual attention away from the road is a critical indicator of accident risk. Such work would suggest that a real-time estimation of driver’s gaze could be coupled with an alerting system to enhance safety when the driver is overly distracted or inattentive High precision eye tracking that includes an estimate of pupil orientation in the vehicle is costly and difficult. From an image processing perspective alone, difficulties involve the unpredictability of the environment, presence of sunglasses occluding the eye, rapid changes in ambient lighting including situations of extreme glare resulting from reflection, partial occlusion of the pupil due to squinting, vehicle vibration, image blur, poor video resolution etc. For example, in, a state-of-the-art algorithm for detecting pupils in the presence of specula reflection achieves only 83% accuracy. In, an accuracy of 87% is achieved for a camera that is positioned off-axis, as it likely may need to be located inside a vehicle. Costs of high resolution recording equipment and other computational requirements further enhance the difficulty of developing practical, deployable solutions. Since pupil detection for eye tracking is often unreliable in real world conditions, the natural question we ask is: how well can we do without it? This is the question that motivated our efforts and makes this work distinct from a large body of literature on gaze estimation. We do not assume that the head pose vector is the same as the gaze vector (i.e., eye pose plus head pose). Most of the automobile accidents are caused by distracted driving. Passively monitoring driver’s eyes can help in detecting state of mind and alertness of driver and thus can reduce risk of accidents. Proposed system includes three main parts 1) Arduino Uno 2) USB web based camera 3) Eyes off the road and fatigue detection