Transformer Model Coupled with Dilated Convolution for Improved Crack Detection

Read full text Summarise with VIRAT
Authors: Joseph James, Lipsa Nayak
Cited by: 0
References: 22

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Structural health monitoring is one of the common and challenging issues in civil engineering.Among the different factors affecting the health of the structure, Crack plays a major role.The identification of cracks is not easy but the early detection of cracks can avoid unpleasant situations in real time.But it exists as a challenge for engineers.This study offers a novel strategy that makes use of deep learning techniques.To get accuracy, the suggested model combines fully linked networks, feature fusion, and feature extraction.To get the best result for feature extraction, a preprocessing module, a dilated convolution network, and a vision transformer are utilized.The model is tested on a dataset that included 40,000 labels labelled" crack and non-crack," and the findings indicate that the suggested model outperforms the state-of-the-art techniques in terms of accuracy, achieving 98.05%.The deep learning technology can help overcome the shortcomings of the most advanced method for categorizing cracks on concrete surfaces, in addition to producing an accurate result.

Fields of Study & Subjects

Infrastructure Maintenance and Monitoring Structural Health Monitoring Techniques Industrial Vision Systems and Defect Detection

Publication details

Journal / source
International journal of intelligent engineering and systems
Volume
19
Issue
4
Pages
988–1003
Publication date
23 March 2026
Language
EN