Analysis of segmentation performance of U-shape neural network algorithm by multi-type encoders for pavement cracks

Pengfei Yong

doi:10.1117/12.2658665

22 December 2022 Analysis of segmentation performance of U-shape neural network algorithm by multi-type encoders for pavement cracks

Pengfei Yong

Author Affiliations +

Proceedings Volume 12460, International Conference on Smart Transportation and City Engineering (STCE 2022); 1246047 (2022) https://doi.org/10.1117/12.2658665
Event: International Conference on Smart Transportation and City Engineering (STCE 2022), 2022, Chongqing, China

Abstract

Current deep convolutional neural networks have achieved suitable applications in the field of crack segmentation. Among them, the decoding-encoding structure network has achieved good results by the long-connected structure for fusing multi-scale features. However, the analysis of the impact of different types of operators in the encoder on the crack segmentation performance needs to be supplemented. This paper uses asymmetric convolution, depth wise separable convolution, and dilated convolution to replace the conventional convolution in the U-shape network, respectively. Moreover, visualization methods are used to explain the extraction characteristics of crack features by different types of operators. The results show that the HDC rule-compliant dilate convolution in the encoding phase of the U-shape network is optimal for the extraction of the crack segmentation, with the MIOU, Recall, and F1 Score improving by 0.0175, 0.0183, and 0.0127 compared to the baseline model consisting of conventional convolution, respectively.

Citation Download Citation

Pengfei Yong "Analysis of segmentation performance of U-shape neural network algorithm by multi-type encoders for pavement cracks", Proc. SPIE 12460, International Conference on Smart Transportation and City Engineering (STCE 2022), 1246047 (22 December 2022); https://doi.org/10.1117/12.2658665

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
6 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Convolution

Computer programming

Feature extraction

Image segmentation

Performance modeling

Visualization

Neural networks

Show All Keywords

Keywords/Phrases

Search In:

Publication Years