A Driver Fatigue Detection Method Based on Eye Aspect Ratio Compensated by Head Pose Estimation

There are many reasons causing traffic accidents, among which fatigue driving is a key factor. By early detection of drivers sleepiness characteristics can effectively reduce the probability of accidents. In this work, we have proposed a non-invasive real-time fatigue detection method based on computer vision, to detect the driver's fatigue using eye aspect ratio. Head pose estimation has been employed to compensate for the error of calculating eye aspect ratio. Experiment results have proved that this method can effectively overcome the errors introduced by the change of head posture to the calculation of eye opening, thus reducing the false alarm rate of driver fatigue detection by 5.09%. The advantages of this system are low hardware cost, non-invasive method, better robustness and high accuracy.