EIDSeg

EIDSeg: A Pixel-Level Semantic Segmentation Dataset for Post-Earthquake Damage Assessment from Social Media Images EIDSeg is a large-scale post-earthquake infrastructure damage segmentation dataset collected from nine major earthquakes (2008–2023).This repository provides the raw dataset in CVAT XML format, along with the corresponding images organized by split.It is intended to be used together with our official codebase for parsing XML annotations and training segmentation models.… See the full description on the dataset page:

Source: Hugging Face Hub (ShyamGhatage/EIDSeg). Metadata imported from the dataset’s Hub tags.