Annotated bridge inspection imagery and videos from AI-powered defect detection in Kansas bridges

This dataset contains annotated images and videos generated from drone-collected bridge inspection imagery used in AI-based defect detection research across four Kansas counties. The files include visual outputs from YOLOv11 instance segmentation models trained to identify and outline eight types of concrete surface defects: Crack, ACrack, Efflorescence, WConccor, Spalling, Wetspot, Rust, and ExposedRebars. Each annotated image shows detection polygons with defect labels and per-image class counts. Supplementary TXT files summarize per-image defect counts. These data were collected as part of research on AI-powered bridge condition assessment conducted at the University of Maryland in collaboration with Kansas counties. The dataset supports reproducibility, benchmarking, and further research on automated infrastructure inspection.