A Deep Learning Based Framework for Distracted Driver Detection

Abstract

AbstractIn this paper, we propose a novel, real-time, deep learning-based framework for distracted driver detection for driver Advanced Driver Assistance Systems (ADAS). We assume that the camera is assumed to be mounted inside the vehicle such that the side view of the driver is in view. Distracted driving is a serious problem leading to a large number of serious and even fatal road accidents worldwide every year. We propose a deep learning architecture that takes as input the captured images of the driver and classifies and recognizes the various distracted driving behaviors. It also recognizes if the driver is not distracted and is alert. The experiments are performed on the publicly available State Farm Distracted Driver Detection (SFDDD) dataset [1] which has 9 classes of distracted driver behavior and one class of alert driving. The training time for the proposed framework is minimal and approach works in real-time. Our experimental results show that our proposed framework is robust and performs better than the state-of-the-art approaches on this dataset.KeywordsDriver assistanceDistracted driver detectionDeep learningDriver behaviorSmart vehicleRoad safety