Automated Damage Prediction for Unreinforced Masonry Walls Using Computer Vision and Symbolic Regression

Seismic events can severely compromise Unreinforced Masonry (URM) walls, heightening collapse risks and escalating repair expenses. Conventional evaluations depend on manual visual inspections, which are inherently subjective and labor-intensive. In contrast, this study introduces an automated methodology that utilizes computer vision to extract critical visual indicators specifically, crack length and crushed area from URM wall images. Employing a broad database of damaged URM wall images, a predictive equation is derived through symbolic regression to link these visual features with the damage state. The model achieves a robust correlation coefficient of ۰.۸۴, highlighting its capability to effectively quantify structural degradation. Overall, this approach offers a fast, precise, and objective tool for post-earthquake damage assessment, thereby enabling more efficient decision-making in repair and reconstruction processes.

کلیدواژه ها:

Unreinforced Masonry (URM) ، Seismic Damage Assessment ، Automated Damage Assessment ، Computer Vision ، Symbolic Regression