Creating xBD: A Dataset for Assessing Building Damage from Satellite Imagery

Ritwik Gupta, Bryce Goodman, Nirav Patel, Ricky Hosfelt, Sandra Sajeev, Eric Heim, Jigar Doshi, Keane Lucas, Howie Choset, Matthew Gaston. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops. [BibTeX]

We present a preliminary report for xBD, a new large-scale dataset for the advancement of change detection and building damage assessment for humanitarian assistance and disaster recovery research. Logistics, resource planning, and damage estimation are difficult tasks after a disaster, and putting first responders into post-disaster situations is dangerous and costly. Using passive methods, such as analysis on satellite imagery, to perform damage assessment saves manpower, lowers risk, and expedites an otherwise dangerous process. xBD provides pre- and post-event multi-band satellite imagery from a variety of disaster events with building polygons, classification labels for damage types, ordinal labels of damage level, and corresponding satellite metadata. Furthermore, the dataset contains bounding boxes and labels for environmental factors such as water, smoke, and lava. xBD will be the largest building damage assessment dataset to date, containing ~700,000 building annotations across over 5,000 km $^2$ of imagery from 15 countries.

Please check out the paper, code, and dataset!