Driver Drowsiness Detection Using Deep Neural Networks

Abstract

Driver drowsiness significantly impacts road safety, leading to numerous accidents. We developed a Driver Drowsiness Detection system using deep learning for binary classification, utilizing CNN, VGG16, GoogleNet, AlexNet, MobileNet_v2, and ResNet101 architectures. Our models were trained on an annotated dataset of driver images and evaluated on metrics like accuracy and F1-score. Results show that while deeper networks offer high accuracy, lightweight models like MobileNet_v2 provide a good balance of performance and computational efficiency. This work demonstrates the potential of these models for real-time drowsiness detection in advanced driver- assistance systems (ADAS) to enhance road safety.