Automatic inspection of cracks on pavement surfaces based on improved segmentation model

Lu Deng; Xu Zhao; Honghu Chu; Chao Xiang

doi:10.1117/12.2658600

22 December 2022 Automatic inspection of cracks on pavement surfaces based on improved segmentation model

Lu Deng, Xu Zhao, Honghu Chu, Chao Xiang

Proceedings Volume 12460, International Conference on Smart Transportation and City Engineering (STCE 2022); 124602U (2022) https://doi.org/10.1117/12.2658600
Event: International Conference on Smart Transportation and City Engineering (STCE 2022), 2022, Chongqing, China

Abstract

Periodic crack inspections are greatly required for the operation and maintenance of the pavement. With the rapid development of the deep learning (DL) method and the computer hardware, manual visual inspections have been gradually replaced by DL-based inspections due to their high efficiency and low cost. In this study, a novel segmentation model was proposed to conduct precise identification of cracks. Firstly, the YOLOv5 was adopted as basic detector for the coarse prediction of the crack type and crack location. Secondly, a mask branch was customized and incorporated into the aforementioned detector for parallel mask production. In addition, the neck, a feature fusion component, was upgraded to produce higher-quality masks, and a sampling-free approach was investigated to reduce class imbalance's influence. Compared to a normal segmentation model, the proposed method demonstrated a better balance between accuracy and speed, with a speed of 33.2 frames per second and an accuracy of 87.13 mAP.

Citation Download Citation

Lu Deng, Xu Zhao, Honghu Chu, and Chao Xiang "Automatic inspection of cracks on pavement surfaces based on improved segmentation model", Proc. SPIE 12460, International Conference on Smart Transportation and City Engineering (STCE 2022), 124602U (22 December 2022); https://doi.org/10.1117/12.2658600

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
8 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Inspection

Image segmentation

Image processing

Performance modeling

Sensors

Convolution

Image processing algorithms and systems

Show All Keywords

Keywords/Phrases

Search In:

Publication Years