Driver Gaze Tracking and Eyes of the Road Detection System

Abstract

Distracted driving is one of the primary causes of Vehicle Accidents in the United States. Inactively checking a driver's activities constitutes the premise of a car security system that can conceivably diminish the quantity of mischances by assessing the driver's concentration of consideration. This paper proposes a cheap vision-based system to precisely distinguish Eyes off the Road (EOR).The system has three primary parts: 1) strong facial element following; 2) head posture and look estimation; and 3) 3-D geometric thinking to recognize EOR. From the video stream of a camera Installed on the guiding wheel section, our system tracks facial components from the driver's face. Utilizing the followed landmarks and a 3-D confront show, the system figures head stance and look course. The head posture estimation calculation is powerful to non inflexible face misshapenings because of changes in looks. At last, utilizing a 3-D geometric investigation, the system dependably recognizes EOR. The proposed system does not require any driver-subordinate alignment or manual introduction and works continuously (25 FPS), amid the day and night. To approve the execution of the system in a genuine auto environment, we led a far reaching experimental assessment under a wide assortment enlightenment conditions, outward appearances, and people. Our system accomplished over 90% EOR exactness for every single tried situation.