Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

xBD: A Dataset for Assessing Building Damage from Satellite Imagery

Ritwik Gupta, Richard Hosfelt, Sandra Sajeev, Nirav Patel, Bryce Goodman, Jigar Doshi, Eric Heim, Howie Choset, Matthew Gaston

TL;DR

This work introduces xBD, a large-scale satellite-imagery dataset for building damage assessment across diverse disasters, integrating pre- and post-disaster imagery, building footprints, ordinal damage labels, and environmental-factor annotations. It defines the Joint Damage Scale to unify labels across disaster types, describes a rigorous annotation pipeline and data collection from the Maxar Open Data Program, and provides dataset statistics, splits, and a baseline model for localization and damage classification. The contributions include a validated multi-disaster damage annotation framework, a substantial annotated corpus (850,736 polygons over ~45,362 km²), and a publicly accessible baseline model to spur development for the xView 2 challenge. The dataset addresses practical humanitarian needs by enabling remote, scalable, and cross-domain damage assessment, potentially reducing on-ground risk and speeding aid delivery.

Abstract

We present xBD, a new, large-scale dataset for the advancement of change detection and building damage assessment for humanitarian assistance and disaster recovery research. Natural disaster response requires an accurate understanding of damaged buildings in an affected region. Current response strategies require in-person damage assessments within 24-48 hours of a disaster. Massive potential exists for using aerial imagery combined with computer vision algorithms to assess damage and reduce the potential danger to human life. In collaboration with multiple disaster response agencies, xBD provides pre- and post-event satellite imagery across a variety of disaster events with building polygons, ordinal labels of damage level, and corresponding satellite metadata. Furthermore, the dataset contains bounding boxes and labels for environmental factors such as fire, water, and smoke. xBD is the largest building damage assessment dataset to date, containing 850,736 building annotations across 45,362 km\textsuperscript{2} of imagery.

xBD: A Dataset for Assessing Building Damage from Satellite Imagery

TL;DR

This work introduces xBD, a large-scale satellite-imagery dataset for building damage assessment across diverse disasters, integrating pre- and post-disaster imagery, building footprints, ordinal damage labels, and environmental-factor annotations. It defines the Joint Damage Scale to unify labels across disaster types, describes a rigorous annotation pipeline and data collection from the Maxar Open Data Program, and provides dataset statistics, splits, and a baseline model for localization and damage classification. The contributions include a validated multi-disaster damage annotation framework, a substantial annotated corpus (850,736 polygons over ~45,362 km²), and a publicly accessible baseline model to spur development for the xView 2 challenge. The dataset addresses practical humanitarian needs by enabling remote, scalable, and cross-domain damage assessment, potentially reducing on-ground risk and speeding aid delivery.

Abstract

We present xBD, a new, large-scale dataset for the advancement of change detection and building damage assessment for humanitarian assistance and disaster recovery research. Natural disaster response requires an accurate understanding of damaged buildings in an affected region. Current response strategies require in-person damage assessments within 24-48 hours of a disaster. Massive potential exists for using aerial imagery combined with computer vision algorithms to assess damage and reduce the potential danger to human life. In collaboration with multiple disaster response agencies, xBD provides pre- and post-event satellite imagery across a variety of disaster events with building polygons, ordinal labels of damage level, and corresponding satellite metadata. Furthermore, the dataset contains bounding boxes and labels for environmental factors such as fire, water, and smoke. xBD is the largest building damage assessment dataset to date, containing 850,736 building annotations across 45,362 km\textsuperscript{2} of imagery.

Paper Structure

This paper contains 31 sections, 10 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Existing Data Sources
  4. Assessment Scales
  5. Requirements for the xBD Dataset
  6. Multiple Levels of Damage
  7. Image Resolution
  8. Diversity of Disasters
  9. Wide Diversity of Buildings
  10. Negative imagery
  11. Joint Damage Scale
  12. Dataset Collection
  13. Imagery Source
  14. Annotation Process
  15. Triage
  16. ...and 16 more sections

Figures (10)

  • Figure 1: From top left (clockwise): Hurricane Harvey; Palu Tsunami; Mexico City Earthquake; Santa Rosa Fire. Imagery from DigitalGlobe.
  • Figure 2: Pre-disaster imagery (top) and post-disaster imagery (bottom). From left to right: Hurricane Harvey; Joplin tornado; Lower Puna volcanic eruption; Sunda Strait tsunami. Imagery from DigitalGlobe.
  • Figure 3: Disaster types and disasters represented in xBD around the world.
  • Figure 4: Joint Damage Scale descriptions on a four-level granularity scheme.
  • Figure 5: Building polygons (shown in green) on imagery from Hurricane Michael (2018)
  • ...and 5 more figures