AI-Based Driver Drowsiness and Distraction Detection in Real-Time

Abstract

This paper proposes a solution to combat risks associated with road accidents namely drowsiness and distractions which have been established to be the prominent causes of accidents worldwide. The proposed methodology integrates camera vision and mathematical computations to accurately detect driver drowsiness and distracted driving. The eye aspect ratio and mouth aspect ratio are utilized to recognize drowsiness characteristics while the eye tracking methodology is adopted to identify distracted behavioral factors. On the detection of the mentioned risk factors, alerts are provided to the driver in visual and audio formats by use of the Raspberry Pi microprocessor, LCD display and buzzer. The developed system was tested under an experimental setup and exposed to various lighting conditions. The results suggested that the approach is capable of recognizing drowsiness and distractions with an accuracy of 94.1% and 89% respectively during both day and night conditions and provide warnings as required.