Realtime Driver Drowsiness Monitoring System using Visual Behaviour and Machine Learning Techniques

Abstract

Fatigue-induced driving is a significant contributor to road accidents and fatalities.
Consequently, there is ongoing research into detecting driver fatigue and signaling its onset.
Traditional approaches predominantly fall into three categories: vehicle-based, behavioral-based,
and physiological-based. Many methods either intrusively disrupt the driver or necessitate costly
sensors and complex data processing. Thus, this study focuses on developing a cost-effective,
real-time drowsiness detection system with satisfactory accuracy. Within the system architecture,
a webcam captures video footage, utilizing image processing techniques to detect the driver's
face in each frame. Facial landmarks are then identified, allowing computation of metrics such as
eye aspect ratio, mouth opening ratio, and nose length ratio. Drowsiness is determined using
adaptive thresholding based on these metrics. Additionally, offline machine learning algorithms
have been integrated. The Support Vector Machine based classification achieves a sensitivity of
95.58% and specificity of 100%..