Automated Drivers Fatigue Detection System Using Machine-Vision Based Technique

Authors

  • Ng Mun Foong , Vikneswary Jayapal , Kalaivani Tarumaraja , Kumaresan Magaswaran , Shamini Pathmanathan

Abstract

Fatigue driving is deliberated a high risk and impact accident aspect on the road, which could
cause injuries and fatalities to the driver. As such, it is very essential to prevent the accidents caused by
fatigued driving all over the world. In this study a non-intrusive system is developed based on machine
vision technique to monitor the state of drivers’ eye and mouth using a camera. Percentage of Eyelid Closure
(PERCLOS) and yawning rates are used to predict the fatigue level. The machine vision-based driver
monitoring system is developed using Histogram Oriented Gradient (HOG) feature descriptor for face
detection and facial points recognition. The algorithm is developed to process the live video feed focused on
the driver’s face and to evaluate the eyes and mouth movements. Then SVM is used to check whether the
detected object is face or non-face. To increase the efficiency of the algorithm, the Eye Aspect Ratio (EAR)
and Mouth Aspect Ratio (MAR) of the driver are implemented to calculate the eye closeness and yawning.
Few practical tests demonstrated that the developed fatigue detection can detect driver fatigue with real time
accuracy of 79.88%.

Published

2020-01-31

Issue

Section

Articles