Driver drowsiness is a major cause of road accidents worldwide, leading to a significant number of fatalities each year. This paper presents a real-time, vision-based drowsiness detection system using Convolutional Neural Networks (CNN). The proposed approach monitors facial expressions and behaviors—such as eye closure and yawning—through a webcam without the need for intrusive sensors. Facial landmarks are detected using OpenCV and Dlib, and the CNN model classifies the driver’s state as alert or drowsy. Preprocessing steps like grayscale normalization and data augmentation improve model robustness across varying lighting conditions. When fatigue is detected, an audible alert is activated to restore driver attention. This study demonstrates the practicality of deep learning in developing cost-efficient, real-time systems for enhancing road safety and supporting intelligent driver-assistance technologies.
Harika Tirumalasetty (Thu,) studied this question.