Driver Drowsiness Detection by Yawn Identification Based on Depth Information and Active Contour Model

Increasing road accidents, caused by sleepy drivers is an important transportation problem now a day. A high percentage of road accidents are caused by drivers' lack of awareness caused by fatigue. Hence, the development of an automated system that determines the driver's drowsiness and issues a warning as soon as the driver loses attention on the road is a necessary step to prevent road accidents. This system must also function under illumination variations that have previously been a complication in similar works. Specific postures of the body and face may be regarded as indications of driver's fatigue. These include yawning and eye closure which determine whether or not the driver is in an adequate condition for driving. In this paper, in order to detect a driver's yawning, her nose tip is first identified in the depth image of the face and then the lower area of the face is separated. In the resulting image, the approximate area of the mouth is found and using depth information and the active contour model, an open mouth is identified. Experiment results show that the presented method can detect yawning with 95% accuracy.