Aerial disaster images captured by the Civil Air Patrol featuring geotagged EXIF metadata and multi-label classification. Each image was labeled in triplicate by trained volunteers following FEMA damage assessment protocols, including infrastructure classes such as 'bridges_any'.
Use Cases
- Train multi-label classification models to detect infrastructure damage using labels like 'bridges_any'
- Perform geospatial mapping of disaster zones by parsing coordinates from the EXIF metadata
- Benchmark damage assessment algorithms against expert-labeled ground truth derived from the FEMA damage assessment process
Strengths
- Aerial disaster images captured and provided by the Civil Air Patrol (CAP)
- Geotagged location information embedded within image EXIF metadata
- Multi-label classification annotations validated through a majority vote of three trained annotators
- Labels aligned with the FEMA damage assessment process for emergency management applications