M. Mudabbir, A. Mosavi, H. Perez: Detecting Building Defects with Deep Learning. EURASIAN JOURNAL OF MATHEMATICAL AND COMPUTER APPLICATIONS Vol. 13, No. 3, pp. 50–67, 2025. ISSN 2306–6172 link

Abstract: Building defects on external walls can include cracks, mould, dampness from waterproofing failures, fungus growth due to high humidity, and paint peeling. These building defects are commonly caused by wear and tear, improper maintenance, and weather conditions. The identification of these defects is very important to maintain the structural health and safety of buildings, which are often a large financial asset. Manual visual inspection is a traditional technique for defect detection and the most laborious way to identify wear defects, in addition to other nondestructive testing procedures that determine defect properties. Advances in DL and computer vision are expected to improve the efficiency of defect detection. For instance, the DL-based YOLOv10 (You Only Look Once) method provides real-time defect detection that is fast and accurate. This study provided the YOLOv10 technique for the automated detection and localization of building defects. In addition, this study not only makes defect detection more efficient but also helps researchers to advance the overall inspection process with more efficiency.