There is a newer version of the record available.

Published October 13, 2026 | Version v0.3.0

mapcv: computer-vision datasets from map imagery and geospatial labels

Authors/Creators

Description

mapcv turns imagery (XYZ tiles, GeoTIFF/COG, Sentinel-2 from STAC or EOPF Zarr, and Google Earth Engine) and geospatial labels (vector files, OpenStreetMap or label rasters) into training datasets for semantic and instance segmentation, object detection, classification, change detection and regression, with georeferenced patches, a manifest and leakage-safe spatial splits.

Notes

If you use mapcv to build a dataset for published work, please cite the version you used (mapcv --version).

Files

tahamukhtar20/mapcv-v0.3.0.zip

Files (3.1 MB)

Name Size
md5:b514e5224bac1498feb9421fc8f03e80
3.1 MB Preview Download

Additional details

Related works