Drowsiness detection based on visual signs: blinking analysis based on high frame rate video

Abstract

In this paper, an algorithm for drivers' drowsiness detection based on visual signs that can be extracted from the analysis of a high frame rate video is presented. A study of different visual features on a consistent database is proposed to evaluate their relevancy to detect drowsiness by data-mining. Then, an algorithm that merges the most relevant blinking features (duration, percentage of eye closure, frequency of the blinks and amplitude-velocity ratio) using fuzzy logic is proposed. This algorithm has been tested on a huge dataset representing 60 hours of driving from 20 different drivers. The main advantage of this algorithm is that it is independent from the driver and it does not need to be tuned. Moreover, it provides good results with more than 80 % of good detections of drowsy states.