software · Global

TerraMind-1.0-base

Hugging Face model ibm-esa-geospatial/TerraMind-1.0-base

Claim this listing

Represent TerraMind-1.0-base? Send a claim for review so we can discuss listing updates. Claims do not automatically grant ownership or verification.

Data offered docs-verified: multimodal Earth Observation foundation model embeddings (per-patch, per-modality) · any-to-any generation of modalities (diffusion-based 'mental images') · Thinking-in-Modalities (TiM) variant with LULC intermediate modality · Sentinel-2 L2A/L1C, Sentinel-1 GRD/RTC, DEM and RGB raw inputs
Formats docs-verified: PyTorch · Hugging Face model weights
Access docs-verified: Open-source model, licence apache-2.0; weights are downloaded automatically from Hugging Face; no API key stated in docs

Use cases llm-inferred: Land-use/land-cover and water body mapping from Sentinel-1/2 for catchment or site baselines · Vegetation assessment from optical imagery as input to habitat (BNG) baseline mapping · Generating missing modalities (e.g. SAR from optical) where one sensor's data is unavailable
Audience llm-inferred: GIS analysts · remote sensing specialists · hydrologists · ecologists · geospatial ML engineers

Constraints llm-inferred: Requires the TerraTorch / PyTorch Lightning Python stack to use · Generated modalities are 'mental images', not reconstructions, so they should not be treated as measured data · Not deployed by any hosted inference provider per the model card · Benchmark claims are from the authors' PANGAEA evaluation; no UK-specific validation is documented

Licence: apache-2.0
Authentication: No API key required (repository README, auto-extracted; check upstream requirements)
GitHub stars: Unknown
Last commit: Unknown
Last successful page check: 14/10/2026
Page health: healthy
Package: Not recorded
Verification: Not independently verified

Verified 2026-10-15 via HF API: exists; 28907 downloads/30d, 54 likes.

Metadata observed 14/10/2026 · research sweep: manual-2026-10-15 (Hugging Face models)