Crack Detection in Concrete Structures Using Deep Learning: A Review

Concrete structures form the backbone of modern infrastructure, but their durability is often compromised by the development of cracks due to environmental stress, loading conditions, and material degradation. Early detection and accurate assessment of cracks are essential for timely maintenance and ensuring structural integrity. Traditional manual inspection techniques, while widely used, are time-consuming, subjective, and prone to inconsistencies. With the advent of artificial intelligence, particularly Deep Learning (DL), there has been a paradigm shift toward automated, accurate, and scalable crack detection systems. This review paper provides a comprehensive analysis of ten recent studies that utilize deep learning for detecting cracks in concrete structures. These studies employ various DL architectures, including Convolutional Neural Networks (CNNs), transfer learning frameworks, ensemble models, and clustering techniques, applied to diverse scenarios such as UAV-assisted inspection and shadow-affected imaging. The review categorizes each approach based on methodology, dataset usage, performance metrics, and practical deployment challenges. Results indicate that DL techniques significantly outperform traditional methods in terms of precision, adaptability, and scalability. However, issues such as domain-specific generalization, lack of standardized datasets, and environmental variability remain critical challenges. The paper concludes by highlighting emerging trends and future research directions that can further improve the reliability and applicability of DL-based crack detection in real-world infrastructure monitoring.

کلیدواژه ها:

Concrete Structures ، Crack Detection ، Image Processing ، Deep Learning ، Convolutional Neural Networks (CNNs)