Real-Time Vision-Based Driver Drowsiness/Fatigue Detection System

In this paper, a vision system for monitoring driver´s vigilance is presented. The level of vigilance is determined by integrating a number of facial parameters. In order to estimate these parameters, the facial features of eyes, mouth and head are first located in the input video sequence. The located facial features are then tracked over the subsequent images. Facial parameters are estimated during facial feature tracking. The estimated parametric values are collected and analyzed every fixed time interval to provide a real-time vigilance level of the driver. A series of experiments on real sequences were demonstrated to reveal the feasibility of the proposed system.