Deakin University
Browse

File(s) under permanent embargo

Vision-based pavement marking detection – a case study

conference contribution
posted on 2021-01-01, 00:00 authored by S Xu, Jun WangJun Wang, P Wu, W Shou, T Fang, X Wang
Pavement markings take responsibility to communicate with road users regarding travel regulations and guidance. Due to their irreplaceable role in ensuring the safety and order on road, it would be beneficial for road agencies to keep an as-is inventory record of the pavement markings on their roads for managerial operations. However, faced with the sheer volume of their responsible assets, manual inspection would be time-consuming and error prone. This study proposes a vision-based method to automatically detect and classify longitudinal markings using videos of road pavement. Not only line markings, audible markings, as a special category, were also identified in the images. The proposed method can achieve an average 0.89 detection accuracy for line markings and 0.82 for audible markings. Limitations and future work are also proposed. This study tests the possibility of utilising visual data to assist road agencies with an informative management of their civil assets.

History

Event

International Society for Computing in Civil and Building Engineering. International Conference (18th : 2020 : Online, Brazil)

Volume

98

Series

International Society for Computing in Civil and Building Engineering International Conference

Pagination

1162 - 1171

Publisher

Springer

Location

Online, Brazil

Place of publication

Cham, Switzerland

Start date

2020-08-18

End date

2020-08-20

ISSN

2366-2557

eISSN

2366-2565

ISBN-13

978-3-030-51295-8

Language

eng

Publication classification

E1 Full written paper - refereed

Editor/Contributor(s)

E Toledo Santos, S Scheer

Title of proceedings

ICCCBE 2020 : Proceedings of the 18th International Conference on Computing in Civil and Building Engineering

Usage metrics

    Research Publications

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC