Using Deep Learning to Construct a Real-Time Road Safety Model; Modelling the Personal Attributes for Cyclist

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper is concerned with the modelling of cyclist road traffic crashes by considering the personal attributes, i.e. gender and age of the cyclists. There are 21 different types of variables considered for each crash, which broadly fall into spatial, infrastructure, and environment categories. The study area of Tyne and Wear county in the north-east of England is selected for investigation. Six deep learning-based safety models are constructed using historic crash data. The effectiveness of deep learning methodology for road safety analysis is demonstrated, and it is found that spatial, infrastructural, and environmental conditions affect the safety interactions of a particular cyclist. These variables can be used for determining/predicting safety for a rider at a location. The model can predict age and gender of the rider, which is likely to be the most unsafe based upon the specific input variables. The significant accuracy is obtained for the constructed models with an overall accuracy of 84%. It is hoped that the proposed models can help in better designing of cyclist network, design, and planning, which will contribute to a sustainable transportation system.
Original languageEnglish
Title of host publicationIntelligent Data Engineering and Automated Learning
Subtitle of host publicationIDEAL 2020
EditorsCesar Analide, Paulo Novais, David Camacho, Hujun Yin
Place of PublicationCham
PublisherSpringer
Pages610-619
Number of pages10
Volume12490
ISBN (Electronic)9783030623654
ISBN (Print)9783030623647
DOIs
Publication statusPublished - Nov 2020
EventIDEAL 2020: 21st International Conference on Intelligent Data Engineering and Automated Learning - Hotel de Guimarães, Guimarães, Portugal
Duration: 4 Nov 20206 Nov 2020
http://islab.di.uminho.pt/ideal2020/

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Number2
Volume12490
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceIDEAL 2020
CountryPortugal
CityGuimarães
Period4/11/206/11/20
Internet address

Fingerprint Dive into the research topics of 'Using Deep Learning to Construct a Real-Time Road Safety Model; Modelling the Personal Attributes for Cyclist'. Together they form a unique fingerprint.

Cite this