Driver Drowsiness Detection: Based on Eye Tracking and Yawning

Abstract

Driver Drowsiness is the fundamental element in road accidents. Around 20% of road accidents are caused by driver’s fatigue during driving, which can be reduced by applying different techniques. Our application provides ways to detect and track drowsiness of the drivers and aware the driver of its drowsiness. Drivers and transportation industries are the main targeted audience of our application. The application detects drowsiness based on various features of the face that are the face, the eyes and the mouth of the driver. These three approaches are proposed in this paper for detection of drowsiness, the head position, the blinking of eyes and the mouth states which consider yawning and normal talking states and differentiates between them. The head position, blinking of eyes and the yawning states are first detected using various algorithms which are tested and practiced on various subjects. The application detects for normal states and abnormal states which includes driver being in the drowsy state by continuously monitoring the face features of the driver. The three different approaches work separately and their results are combined at the end and a value is obtained. A threshold has been fixed and if the value exceeds the threshold, an alarm is generated. Hence in this paper, a measure is taken to reduce as many accidents as possible that are caused by the driver’s drowsiness