Safety Enhancement In Vehicles: Drowsiness Detection Using Convolutional Neural Networks

Abstract

Advancements in computer vision have significantly
benefited drivers, particularly through technologies
like self-driving cars. Approximately 20% of
accidents result from driver fatigue and drowsiness,
posing a serious issue. While numerous solutions
have been proposed to address this, many are
unsuitable for real-time application. These methods
struggle with challenges such as variations in facial
features and changes in lighting conditions. Our goal
is to implement an intelligent system that significantly
reduces road accidents by monitoring key facial
features of the driver, such as eye closure, blinking
rate, yawning, and head movements. This system
continuously observes the driver using a webcam.
Facial features, including eye movements, are
tracked using a cascade classifier. Images of the eyes
are extracted and processed through a customdesigned
Convolutional Neural Network (CNN) to
determine whether both eyes are closed. Based on this
classification, an eye closure score is generated, and
if the system detects signs of drowsiness, an alarm is
triggered to alert the driver.