Detection Models for Concrete Crack Using Deep Learning Approaches: A Review

AYDIN Y., ORMAN Z., BEKDAŞ G.

Structural Engineering International, 2026 (SCI-Expanded, Scopus) identifier identifier

  • Publication Type: Article / Article
  • Publication Date: 2026
  • Doi Number: 10.1080/10168664.2025.2576470
  • Journal Name: Structural Engineering International
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, ICONDA Bibliographic
  • Keywords: artificial intelligence, CNN, concrete, crack detection, deep convolutional neural networks, deep learning
  • Istanbul University Affiliated: No

Abstract

The detection and examination of cracks in concrete is important for structural health control and requires time and money. Therefore, new approaches are being sought for crack detection. In this article, studies in the literature on the use of deep learning (DL) in crack detection are reviewed. The study mainly focuses on studies between 2018 and 2024 in order to present the latest scientific developments on crack detection. These studies were reviewed in terms of the datasets and performance metrics used. The main objective of this study is to identify, evaluate, and analyze convolutional neural network (CNN) models used in crack detection applications. The review was conducted based on two main criteria: first, the CNN architectures used for classification; second, the accuracy rates achieved by these models in each dataset. The findings reveal that CNN-based approaches have significant potential in crack detection.