Quantcast

An Efficient Approach to Detect Driver Distraction during Mobile Phone Usage

Research paper by Medha Sawhney, Vasundhara Acharya, Krishna Prakasha

Indexed on: 23 Dec '18Published on: 19 Dec '18Published in: International Journal of Engineering & Technology



Abstract

A distracted driver is an invitation to a fatal vehicle accident. People lose their lives every day due to distracted driving and using mobile phones while driving is one of the primary reasons behind road accidents. Hence, detection of mobile phone usage to alert the driver or for an autonomous system to take over becomes extremely important. In an attempt to solve this issue of distracted driving, the authors proposed a Convolution Neural Network (CNN)  based model to detect mobile phone usage by the driver. The proposed work presents not only a practical solution to the problem but also a comparison between traditional approaches (Support Vector Machine with HOG) and a CNN based model. The traditional methods are both implemented and tested by the authors. The presented model performs input segmentation to achieve an efficient accuracy of 97%. Deep learning was found to be the best solution to detect driver distraction while on a call accurately