A multi-class driver behavior dataset for real-time detection and road safety enhancement

Abstract

This paper introduces a novel dataset designed to support the development of AI-driven driver monitoring systems. The dataset captures real-world driver behaviors under diverse driving conditions, including private vehicles and public buses, in Dhaka, Bangladesh. It comprises 7286 high-resolution images categorized into five behavioral classes: Safe Driving, Talking on the Phone, Texting, Turning, and Other Distracting Behaviors. The dataset reflects natural variations in driver behavior, such as different lighting conditions, angles, and vehicle types, making it highly applicable to real-world scenarios. By providing a comprehensive and annotated dataset, we aim to support the development of intelligent transportation systems and contribute to reducing accidents caused by distracted driving. The dataset is publicly available and can be used to train and evaluate machine learning models for real-time driver behavior detection.