Detection of Drowsiness in Drivers Using TensorFlow Framework

Abstract

Drowsiness among drivers is increasingly contributing to a significant number of accidents on roadways, emerging as a leading cause of such incidents. Recent findings highlight the substantial impact of driver fatigue on vehicular accidents, resulting in numerous fatalities. In fact, tiredness is responsible for more than 30% of all accidents. To address this issue and potentially save lives, a framework is proposed for detecting driver sleepiness and alerting the driver. The framework involves continuous monitoring of the driver through a camera, employing image-processing algorithms that focus primarily on the driver's face and eyes. By analyzing the location of the driver's eyes, the model predicts eye blinking patterns. To calculate PERCLOS (percentage of eye closure), an algorithm tracks and analyses the driver's face and eyes. If the driver's squinting rate exceeds a certain threshold, the framework triggers an audible alert.