Determination Of Structural Cracks Using Deep Learning Frameworks

2025-07-03

AI Summary

To address the limitations of conventional manual crack detection—including low speed, subjectivity, and high miss-rate—this paper proposes a multi-model ensemble architecture integrating Residual U-Net, SegNet, and U-Net, augmented by a novel Convolutional Meta-Model for feature-level fusion. The framework exhibits high robustness and precise localization even on low-resolution input images. Evaluated on public crack datasets, the ensemble achieves an IoU of 89.7% and a Dice coefficient of 94.2%, significantly outperforming individual constituent models and state-of-the-art methods. Experimental results demonstrate that the proposed approach effectively balances detection accuracy and computational efficiency. It thus provides a reliable, deployable deep learning solution for automated infrastructure inspection—particularly in critical applications such as bridge and tunnel monitoring.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Ensemble Methods
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Structural crack detection is a critical task for public safety as it helps in preventing potential structural failures that could endanger lives. Manual detection by inexperienced personnel can be slow, inconsistent, and prone to human error, which may compromise the reliability of assessments. The current study addresses these challenges by introducing a novel deep-learning architecture designed to enhance the accuracy and efficiency of structural crack detection. In this research, various configurations of residual U-Net models were utilized. These models, due to their robustness in capturing fine details, were further integrated into an ensemble with a meta-model comprising convolutional blocks. This unique combination aimed to boost prediction efficiency beyond what individual models could achieve. The ensemble's performance was evaluated against well-established architectures such as SegNet and the traditional U-Net. Results demonstrated that the residual U-Net models outperformed their predecessors, particularly with low-resolution imagery, and the ensemble model exceeded the performance of individual models, proving it as the most effective. The assessment was based on the Intersection over Union (IoU) metric and DICE coefficient. The ensemble model achieved the highest scores, signifying superior accuracy. This advancement suggests way for more reliable automated systems in structural defects monitoring tasks.

Problem

• Automating structural crack detection to improve accuracy
• Overcoming limitations of manual inspection methods
• Enhancing efficiency with deep learning ensemble models

Innovation

• Novel deep-learning architecture for crack detection
• Residual U-Net models integrated with meta-model
• Ensemble model outperforms SegNet and U-Net