Titlebook: Artificial Intelligence in Vision-Based Structural Health Monitoring; Khalid M. Mosalam,Yuqing Gao Book 2024 The Editor(s) (if applicable)

Muscularis

Structural Damage Localizationbased SHM. As introduced in Chap.?., damage locations are represented by bounding boxes?with coordinates and localization algorithms and models aim to conduct a regression task to accurately predict the coordinates, which closely match the ground truth.

鈍劍

Active Learninga. If the SSL method is employed, both labeled and unlabeled data can be utilized simultaneously, which can improve the AI model performance to some extent, as demonstrated with the BSS-GAN in Chap.?..

落葉劑

Multi-task Learning. However, many previous studies solely work on the existence of damage in the images and directly treat the problem as a single attribute classification or separately focus on finding the location or area of the damage as a localization or segmentation problem. Abundant information in the images fr