Title: Distracted driver recognition system using deep forest
Authors: Moolchand Sharma; Ananya Bansal; Utkarsh Agrawal; Pramod Goyal
Addresses: Maharaja Agrasen Institute of Technology, Delhi, India ' Maharaja Agrasen Institute of Technology, Delhi, India ' Maharaja Agrasen Institute of Technology, Delhi, India ' Maharaja Agrasen Institute of Technology, Delhi, India
Abstract: The World Health Organization (WHO) reported around 1.25 million yearly deaths due to road traffic accidents, and the number has been continuously increasing over the last few years. We need to develop an accurate system for detecting distracted drivers and warn them against it. Distracted driving is an ongoing problem that seems only to be getting worse with the dependence on technology. We have used a novel approach, i.e., deep forest, which is a recently researched algorithm, developed as an alternative to deep learning models based on neural networks. We aim to implement deep forests for classification and to compare the results with other techniques. Experimental results shows that the system outperforms well, and achieving an accuracy of 96.75 with DeepForest. The success of the models will, hopefully, one day, aid combat the ongoing and increasing issue of distracted drivers on the roads.
Keywords: convolution neural network; deep forest; decision trees; ensemble-based; non-differentiable; deep learning.
International Journal of Electronic Business, 2021 Vol.16 No.4, pp.336 - 351
Received: 30 Mar 2020
Accepted: 06 Feb 2021
Published online: 27 Oct 2021 *