Enhanced Driver Vigilance System

Abstract: The concerns rise in driver fatigue-related vehicle collisions has made drowsiness detection in drivers a significant area of study. Experts say that drivers who drive long distance without taking regular rests are at a high risk of experiencing fatigue. Research shows that exhausted drivers in need of rest account for around 25% of all major highway collisions. The purpose of our systems is to spot early indicators of driver exhaustion before they impact one’s ability to drive. This system is a novel approach utilizing deep learning techniques, specifically 2D convolutional neural networks (CNNs), to identify signs of drowsiness in drivers face by analysing facial and eye features. The idea is aimed to use traditional models of multi-layer 2D CNN with multi-label classification and Haar-cascade algorithm. Multiple face signs like eye closures and yawning are considered through the input images to improve the detection accuracy under various driving conditions.

Keywords: Drowsiness detection, Facial recognition, eye detection, yawn detection, multi-label classification, Image-based analysis, Deep learning.