Real-Time Driver Distraction Detection Using Fast R-CNN Algorithm

Driver distraction is a leading cause of road accidents, highlighting the need for detection systems. This article presents a new real-time driver distraction detection system based on Fast R-CNN in real-time with the highest accuracy of 89%. Unlike previous studies, our solutions involve the CNN and the transfer learning of electrode arrangement and feature extraction for identifying activities like yawning, averting one's gaze, and using the phone. Our system also includes a temporal analysis module that uses long short-term memory to capture temporal relations in driver behavior. This approach provided a novel way of identifying drivers' distractions autonomously, leading to a 30% reduction in accidents. This research enriches the knowledge base for intelligent driver assistance systems, leading to improved road safety levels worldwide.