Vision-Based Driver's Attention Monitoring System for Smart Vehicles

Recent studies revealed that the driver's inattention is one of the most prominent reasons for car accidents. Intelligent driving assistant system with real time monitoring of the driver's attentional status may reduce the accident rate that mostly occurred due to lack of attention. In this paper, we presents a vision-based driver's attention monitoring system that estimates the driver's attentional status in terms of four categories: attentive, distracted, drowsy, and fatigue respectively. The attentional status is classified with a variety of parameters such as, percentage of eyelid closure over time (PERCLOS), yawn frequency and gaze direction. Experimental results with different subjects show that the system can classify the driver's attentional status with a reasonable accuracy.