Driver Fatigue and Drowsiness Detection

Based on Facial Features Monitoring

 

 

 Motivation

 

According to the World Health Organization (WHO), it is anticipated that road traffic accident will be the third leading cause of deaths and injuries in the world, by the year 2020.

 

Statistics show an annual rate of 1.3 million deaths and 50 million injuries for traffic accidents, worldwide, and the global cost of traffic crashes is estimated at $518 billion per annum.

 

US National Highway Traffic Safety Administration (NHTSA) reveals that driver fatigue and distraction is the leading factor in 90% of the road accidents. Crashes due to distraction, or drowsiness are usually lead to the most catastrophic and severe accidents, as there is no speed reduction and no control on vehicle dynamics (a drunken driver still has some levels of control). Therefore, introducing a preventive safety system that is able to evaluate the driver's level of alertness as well as preventing imminent crashes is a highly interesting topic not only in academia, but also for automotive industry, and traffic authorities.

 

 

Proposed Work

 

Utilizing techniques of computer vision and multi-senor data fusion, our research proposal aims to monitor both visual and cognitive signs of driver distraction while driving, in real-time. The final outcome of the research project will be to detect driver drowsiness, distraction and fatigue via real time processing of images which are captured by multiple cameras mounted as below figures for both on-board and out-road environment monitoring. Although many work has already done on drowsiness detection via eyelid monitoring, a very limited known studies have considered to measure “driver cognitive fatigue” based on camera data. Most of the cognetive fatigue measurement methods rely on EEG, ECG and other biometric means. Our Computer-vision based method assesses the driver’s level of fatigue, drowsiness and distraction based on a multi-clue fusion techniques, with the final objective of preventing imminent crashes due to driver's innatention.