A DEEP LEARNING BASED SLEEPNESS AND WAKEFULNESS DETECTION FOR DRIVERS

Abstract

Falling asleep while driving is a major part of road accidents. Traffic accidents can be considered as a public health problem and several factors like drugs, driving without rest, sleep disorders, alcohol consumption affect sleep deprivation. Furthermore, drivers are also unaware of falling asleep situations, such as highway hypnosis. All these factors cause accidents while driving and are often fatal. A good background should be provided for drivers to implement effective driver warning systems and other countermeasures just before the accident. In this study, Long-Short Term Memory (LSTM) based driver warning system has been proposed to prevent road accidents. The Electrocardiogram (ECG) signals are processed instantaneously to check whether they go into sleep or not. Experimental studies have been carried out on two different human data sets as sleep mode and awake mode. The %95.52 accuracy rate confirms the effectiveness of the proposed method and show its superiority over some state-of-the art methods.

Keywords

• deep learning
• driver sleepiness detection
• electrocardiogram
• staying awake
• driving