Evaluating In-Domain Transfer Learning for Multispectral Land Cover Mapping in Low-Data Regimes - Supplementary material part II

This deposit contains additional supplementary data related to the publication "Evaluating In-Domain Transfer Learning for Multispectral Land Cover Mapping in Low-Data Regimes" (DOI: 10.5281/zenodo.14163353 to main Zenodo article supplementary deposit). Ths submitted data contains OpenStreetMap (OSM [*]) class files originally downloaded from Geofabrik, and contains PBF and Geopackage files (OSM_data.zip) to complete the making of OSM class TIFFs used to train deep learning models for land cover and land use classification together with Sentinel-2 optical images. The ne_10m_ocean.zip contains additional OSM ocean classes downloaded from Natural Earth, to add OSM classes for ocean and sea water near land where OSM lacks ocean coverage. These additional data are added in this new deposit due to large file sizes. ------------------- OSM_datazip - Data from Geofabrik: - PBF: - europe-latest.osm.pbf (07.0525) - great-britain-latest.osm.pbf (21.05.25) - south-korea-latest.osm.pbf (21.05.259 - Geopackage: - australia-latest.osm.pbf (08.05.25) - germany-latest.osm.pbf (09.05.25)ne_10m_ocean.zip - Data from Natural Earth: - ne_10m_ocean.shp (++) (08.05.25) [*] OpenStreetMap is licensed under the Open Database License (DObL) (openstreetmap.org/copyright).