Deep Learning-based Pavement Cracks Detection via Wireless Visible Light Camera-based Network

Yuxin Zou, Wen Cao, Mingyuan Luo, Peng Zhang, Wei Wang, Wei Huang

科研成果: 书/报告/会议事项章节会议稿件同行评审

3 引用 (Scopus)

摘要

Road maintenance is important to ensure the normal usage of road transportation. Due to complicated effects including natural consumptions and human overuses, various forms of damages appear on the road surface, and among them, pavement cracks are the most representative ones. Therefore, the automatic detection of pavement cracks is important and essential in road maintenance. In this study, deep learning techniques are incorporated to fulfill the automatic pavement cracks detection based on a wireless visible light camera-based network. Technically, a visible light high-definition camera is equipped at the bottom of a maintenance vehicle, and pavement images are automatically captured along with the movement of the vehicle. Then, images are wirelessly transferred to a workstation on the maintenance vehicle, and the pavement cracks detection task is fulfilled via deep learning techniques from the classification perspective. Experimental results demonstrate that, the above wireless visible light camera-based system is capable to discern pavement cracks accurately. Also, real-time performance is guaranteed within this deep learning-based pavement cracks detection study.

源语言英语
主期刊名2019 Computing, Communications and IoT Applications, ComComAp 2019
出版商Institute of Electrical and Electronics Engineers Inc.
47-52
页数6
ISBN(电子版)9781728119731
DOI
出版状态已出版 - 10月 2019
活动2019 IEEE International Conference on Computing, Communications and IoT Applications, ComComAp 2019 - Shenzhen, 中国
期限: 26 10月 201928 10月 2019

出版系列

姓名2019 Computing, Communications and IoT Applications, ComComAp 2019

会议

会议2019 IEEE International Conference on Computing, Communications and IoT Applications, ComComAp 2019
国家/地区中国
Shenzhen
时期26/10/1928/10/19

指纹

探究 'Deep Learning-based Pavement Cracks Detection via Wireless Visible Light Camera-based Network' 的科研主题。它们共同构成独一无二的指纹。

引用此