Driver Drowsiness Detection Using Facial Landmark Analysis Technique

Abstract

Drowsy driving is considered to be one of the major reason for road accidents. Lives of the pedestrians, fellow passengers, as well as neighboring travelers are put to risk when driver tend to fall asleep on the steering wheel while driving. In our proposal, early drowsiness of the driver is detected using computer vision techniques to detect the facial landmarks of the driver’s face and the eye blinking patterns are computed using EAR (Eye Aspect Ratio) formula. First, the facial area of the person is detected and the coordinate points for both left and right eye are obtained. Then, EAR is calculated using these sets of coordinates. Next, this EAR is compared with a threshold to check whether the driver is drowsy or not and accordingly alarm is beeped. Instead of using traditional image processing technique to detect eye closure where it checks for disappearance of the white region of the eyes for a period of time, we use a mathematical formula to detect eye closure and it is more convenient and accurate.