VIGILANT VISION: HARNESSING FACIAL ANALYSIS FOR ROAD SAFETY AGAINST DROWSY DRIVING

Abstract

Drowsy driving poses a critical challenge in India, contributing significantly to road accidents and fatalities. This research explores a solution leveraging facial recognition tech- nology, specifically the VGG Face 16 architecture, for drowsy driving detection. The study aims to develop a robust system capable of early identification of drowsiness cues in drivers, potentially preventing accidents. The proposed solution outlines the architecture, dataset description, preprocessing steps, and working model of the VGG16-based system. The study’s working process involves model training, validation, and visualization of performance metrics.The study’s methodology entails model training, performance metric visualization, and assessment. The trained model exhibits a notable accuracy of 97.35 percent during the training phase. Further evaluations and testing on unseen data are advised to validate the model’s real-world effectiveness and generalization. This research seeks to contribute significantly to mitigating the impact of drowsy driving on road safety and public well-being in India.

Index Terms—Drowsy Driving, Facial Recognition, VGG Face Model, Driver Drowsiness Detection, Road Safety, Deep Learning