software · Global

satlas-pretrain

Hugging Face model allenai/satlas-pretrain

Claim this listing

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

Data offered docs-verified: Pre-trained Swin Transformer and ResNet backbones for remote sensing imagery · Backbones pre-trained on Sentinel-2, Sentinel-1, Landsat 8/9 and 0.5-2 m/pixel aerial imagery · SatlasPretrain dataset: remote sensing images paired with labels from OpenStreetMap, WorldCover and manual annotation
Coverage llm-inferred: Global
Resolution docs-verified: 0.5-2 m/pixel for aerial imagery; satellite resolution not stated
Formats docs-verified: tar archives (dataset) · PyTorch .pth (model weights)
Access docs-verified: Free open download from Hugging Face. Model weights released under ODC-BY per the body text, but the page header shows apache-2.0 (ENTRY also says apache-2.0). Conflicting; verify before commercial use.

Use cases llm-inferred: Fine-tune a pre-trained backbone on local aerial or satellite imagery for land-cover or feature detection in an environmental baseline · Detect tree cover or infrastructure features (e.g. solar farms, wind turbines) as inputs to a habitat or planning screening study · Build an in-house imagery classifier for catchment or site land-use mapping in a civils or drainage project
Audience llm-inferred: GIS analysts · remote sensing specialists · ecologists · planners · machine learning engineers

Constraints llm-inferred: Provides pre-trained model weights and training data, not a ready-to-use data API or UK-specific product · Needs PyTorch/torchvision and fine-tuning work to apply to a downstream task · Licence conflict between page header (apache-2.0) and body text (ODC-BY) · No update frequency or dataset time span stated in the docs · Input normalisation and usage details are referred to the GitHub repository and are not in the docs text · No inference provider deployment listed; downloads not tracked

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; 0 downloads/30d, 22 likes.

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