Radiant MLHub: Geospatial Training Data for Earth Observation Machine Learning

Radiant MLHub is an open library for geospatial training data hosted by the Radiant Earth Foundation. It aggregates datasets contributed by the foundation's team and its partners, stored using the SpatioTemporal Asset Catalog (STAC) standard and exposed via a common API. The datasets include pairs of imagery and labels for machine learning tasks like image classification, object detection, and semantic segmentation.

Use Cases

• Train image classification models for land cover mapping based on labeled satellite imagery.
• Develop object detection models for identifying infrastructure or natural features based on annotated imagery.
• Perform semantic segmentation for detailed environmental analysis based on pixel-level labels.
• Benchmark geospatial machine learning algorithms using standardized, STAC-compliant training data.

Strengths

• Data is stored and cataloged using the open SpatioTemporal Asset Catalog (STAC) standard.
• Labels are generated from ground reference data and/or image annotation, as stated in the description.
• Datasets are contributed by multiple partners, suggesting a collaborative sourcing method.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Radiant Earth Foundation and its partners

Collection Method

Datasets generated by the foundation's team and contributed by partners.

Time Range

unknown

Freshness

Last updated: unknown

Geography

unknown