Sat Bbox Metadata Sft V1: Satellite Imagery Chips with Captions and Bounding Boxes

NuTonic/sat-bbox-metadata-sft-v1 is a metadata-first dataset built for training multimodal chat models. It likely contains Sentinel-2 satellite image chips paired with JSON metadata and optionally Mapbox stills. The dataset was created by NuTonic and last updated on April 28, 2026.

Use Cases

• Captioning satellite imagery chips based on the description mentioning 'Captioning for satellite chips'
• Grounding land-cover regions via bounding boxes based on the description mentioning 'Grounding (bounding boxes in normalized coordinates) for land-cover regions'
• Generating class-focused captions based on the description mentioning 'Class-focused captions'
• Performing absence checks for specific land-cover classes based on the description mentioning 'absence checks'

Strengths

• Dataset is designed for 'high-signal, production-shaped supervision' for multimodal models
• Built from an existing 'sat-bbox' style dataset tree, suggesting a structured foundation
• Last updated on April 28, 2026, indicating recent maintenance

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Description metadata is limited; actual data quality requires manual inspection after download

Provenance

Source

Built from an existing 'sat-bbox' style dataset tree (likely Sentinel-2 chips)

Collection Method

Procedural dataset built from satellite chips and JSON metadata sidecars

Freshness

Last updated 2026-04-28 13:30:41; freshness should be verified