IDEA: Image Database for Earthquake damage Annotation
Authors/Creators
Description
The “Image Database for Earthquake damage Annotation (IDEA)” is an extended dataset of annotated images of real-world structures with damage.
The dataset contains images collected during post-earthquake reconnaissance surveys or ordinary inspections, and subsequently annotated by structural engineers, covering different structural typologies, construction materials and damage mechanisms.
The dataset is annotated referring to a refined ontology for structural elements and damage typolgies. A secondary, broader classification of damage includes the following: Crack, Spalling, Corrosion/Exposed rebar, Leaching, Debris, Other damage/mechanism, Non structural damage.
IDEA aims to provide data for the development deep learning-based methods for automated structural damage detection or for training and dissemination purposes.
Technical info
The dataset is constituted by two subsets, depending on whether in each image there is presence (“damage” folder) or absence (“no_damage” folder) of any type of structural damage. Each subset contains two subdirectories, namely “images” and “annotations” that are consisting of image files (.jpg) and relative annotation file (.xml), respectively.
The subdirectories “annotations” contain the XML files in PASCAL VOC corresponding to each image. Every XML file contains information on folder, filename, damagePresence, structureType (typology of structure), size (height and width of the image), object type (type of structural element affected by damage), object type including the type of annotation box (rectangular or polygonal) and damage typology.
Notes
Files
damage.zip
Additional details
Related works
- Is cited by
- Journal article: 10.1016/j.engappai.2025.110302 (DOI)
- Conference paper: arXiv:2412.05042 (arXiv)
- Is described by
- Journal article: 10.1016/j.dib.2025.111733 (DOI)
Dates
- Available
-
2025