Distracted Driver Detection Using Deep Learning Models

Abstract

A leading cause for road accidents is distracted driving. To design an efficient driver assistance system in order to reduce road accidents, there is a need to detect distractions while driving. Impressed by the Deep Learning Neural Network performance in image classification (in Computer Vision), this paper focuses on detecting distracted driving. In this work, authors implemented AlexNet, VGG-16 and GoogLeNet (Inception V1) on State Farm Distracted Driving dataset. Results have shown that GoogLeNet model which has an accuracy of 98.94% outperforms the previous approaches of distracted driver detection by more than 10% on State farm distracted driving dataset.