Validated damage areas

Venezuela M7.5 earthquake, 24 Jun 2026. AI-detected, MapSwipe-validated.

Method

Building damage assessment working flow

The SAR (Sentinel-1 radar) source shown above is not part of this validation, since radar change is hard to confirm by visual inspection.

HOT Activation Area

Method

Field validation working flow

Data schema

The file behind the Remote Validated map.

validated/venezuela_validated_damage.geojson, CRS EPSG:4326, geometry MultiPolygon (H3 resolution-11 cells). Full dataset schema: SCHEMAS.md.

How the verdict is decided

Each area is shown to several MapSwipe volunteers, who mark it damaged or not sure. damaged_share is damaged / total_validators.

A majority, more than 50 percent, marking it damaged makes the verdict accepted (confirmed damage). Anything at or below 50 percent stays uncertain.

Fields

FieldTypeMeaning
areastrCity slug: la_guaira, catia_la_mar, caraballeda, naiguata, caracas, moron, el_junquito, carayaca. Called area here, city in the rest of the dataset.
project_idstrMapSwipe project id.
project_namestrMapSwipe project name.
task_idstrMapSwipe task id for the area.
verdictstraccepted (confirmed damage) or uncertain (not crowd-confirmed).
answerstrMajority answer: damaged or not sure.
damagedintVotes marking the area damaged.
not_sureintNot-sure votes, including reviewers who did not mark damage.
total_validatorsintNumber of volunteers who voted on the area.
damaged_sharefloatdamaged / total_validators, 0.0 to 1.0.
sourcesstrWhich AI source flagged the area for validation: fair, microsoft, or fair,microsoft.
imagery_tmsstrTMS URL of the imagery shown to volunteers.