Evaluating Visual Damage Features for Predicting Damage in Unreinforced Masonry Walls Using Machine Learning

This study explores the use of Machine Learning (ML) to assess the damage state of Unreinforced Masonry (URM) walls based on visual damage features. Earthquakes often result in visible surface damage, such as cracks and crushed areas, which can provide critical insights into the structural condition. In this research, key visual indicators-crack length and crushed area-were extracted using computer vision techniques. These features were then used as input for machine learning models to predict the damage state of URM walls. The results demonstrate that visual damage features contain meaningful and sufficient information for accurately predicting structural damage. This finding highlights the potential of integrating visual data and ML algorithms into automated post-earthquake assessment systems, enabling faster and more reliable decision-making for repair and reconstruction efforts.

کلیدواژه ها:

surface damage ، unreinforced masonry wall ، damage assessment ، computer vision ، machine learning