Automated Crack Analysis Using Deep Learning for Concrete Structures

Volume Title: ICASGE2025

Paper ID : 1042-ICASGE-FULL

The American University in Cairo

Cracks in concrete structures are essential indicators of potential structural degradation, emphasizing the importance of precise crack detection and measurement. This paper introduces a deep learning model designed to detect and segment cracks in concrete surfaces with high accuracy, focusing specifically on measuring crack width, height, and area coverage. These characteristics are critical for assessing the severity and potential impact of structural damage, as larger or more extensive cracks often signal underlying issues that require timely intervention. By leveraging pixel-level semantic segmentation, the model accurately captures crack dimensions, providing a detailed analysis of the crack’s extent and coverage across the concrete surface. This information enables a more thorough understanding of the structural health of concrete, guiding maintenance efforts to areas with significant deterioration. Automating the detection and measurement of cracks through this approach offers a scalable, efficient solution for real-time infrastructure inspection. Ultimately, this method supports proactive infrastructure management and contributes to the long-term safety and resilience of concrete structures.

Computer Vision; Deep learning; Concrete Degradation; Automated Detection and Convolutional Neural Network