AI-based non Destructive Crack Detection in Masonry Structures - 3Pager

The assessment of structural condition in masonry buildings remains a key challenge for enabling circular construction practices, particularly when damage develops progressively and is not easily detectable at early stages. Crack formation is one of the most common indicators of structural degradation in masonry, and its timely detection is essential to prevent further deterioration and support maintenance and reuse decisions. Conventional inspection approaches rely heavily on manual visual assessment, which is often subjective, time-consuming, and difficult to scale across large building stocks.

This work addresses these limitations through the development of an AI-based, non-destructive methodology for crack detection and assessment using image data. The approach combines machine learning models, such as Random Forest and XGBoost, with deep learning architectures including Convolutional Neural
Networks (CNN) and UNet, enabling both classification and pixel-level segmentation of cracks. A custom dataset of masonry surfaces is collected and annotated to support supervised training and evaluation.

Aligned with the REINCARNATE innovation “Building Inspection & Valuation” and the WP1-related tool “Predictive Life-Cycle Information”, the approach improves reliability, scalability, and objectivity of inspections and supports lifecycle-oriented decision-making