Abstract: AbstractThis paper introduces a novel framework that combines vision-language models (VLMs) and localization techniques to detect, classify, and localize visual structural defects using moving platforms such as robots and ...