Enhanced Driver Drowsiness Detection Solution with CNN Algorithm Integration

Abstract

Frequent accidents occur due to drivers experiencing drowsiness, especially during nighttime journeys where fatigue can lead to unintentional sleep. To address this issue, our system employs advanced machine learning techniques, particularly Deep Learning algorithms, to accurately detect signs of driver tiredness. Unlike older methods, our system utilizes Convolutional Neural Networks (CNNs) for enhanced accuracy and adaptability across various environmental conditions. CNNs offer real-time processing, resilience to lighting changes, and the ability to capture intricate spatial features, making them superior to previous image processing techniques. Our model incorporates Haar cascade Classifier for facial feature extraction and Shi-Tomasi corner detection, as well as Lukas-Kanade optical flow algorithm for tracking head movements. With these methods, we achieved a high accuracy rate of 98%.