Development of Doze Detection Method in Real-time Processing Using Smartphone

Abstract

The drowsiness detection system has been developed because the traffic accident caused by the doze during driving does not decrease. The drowsiness detection systems use the various biological information, e.g. heart rate variability and eye movement. These methods are burdensome because they need recording electrodes. The commercially available doze detection system is only available as an option of deluxe car. Thus, the doze detection system still does not become popular. The drowsiness detection system using conventional device in low price is effective to become widely used and reduce traffic accidents.We have developed the doze detection system using only blink information. This system can detect the drowsy state in low arousal level. However the system cannot detect the drowsiness and alarm in real-time because the system operates off-line. In this study, the real-time doze detection algorithm was implemented in a smartphone and the accuracy of doze detection system was evaluated.