A mobile application for driver's drowsiness monitoring based on PERCLOS estimation

Abstract

Vehicular accidents caused by driver drowsiness involve about 7,000 people/year in Brazil, only on federal highways, and cause psychic damage and traumatic stresses for both the victims involved and their families. Drowsiness is characterized by reduced level of vigilance and concentration, which are essential during driving activity. Due to this adversity, many applications of drowsiness detection had been continuously developed through electrical body signals to alert the driver at the time when sleepiness is identified, such as heart rate variability (HRV) and electroencephalogram (EEG). Although these methods work, the use of electrodes in the driver's body is highly invasive. Therefore, we propose a drowsiness detection system based on driver's real time video capture, by estimating the percentage of eyelid closure over a period, without any contact device. Since the use of smartphones has been growing in the last decade, the system has been implemented in a mobile phone even with memory and processing limitations. Processing reduction procedures were developed to improve the application performance, such as the reduction of the region of interest and the limitation of the search window, which increased by 93.09% the number of frames per second and allowed the application to operate smoothly.