Satellite Vision Toolkit
Clear, multi-level interpretation of local urban overhead imagery—from whole-scene LULC context to pixel cover and individual objects.
Urban sample scenes
High-resolution 256×256 USGS aerial chips · click any card to load
Dense residential
Buildings · streets · impervious cover
Urban intersection
Roads · vehicles · pavement
Marina / harbor
Water · boats · harbor context
Parking lot
Pavement · dense small vehicles
Select an urban case to load its acquisition details and analysis prompt.
Run analysis
Upload your own image or start with an urban case. Default settings suit most previews.
Best for: local RGB satellite/aerial chips where buildings, roads, water, or supported objects are visible. Results are model estimates, not surveyed GIS data.
Scene classification probability profile
Ranked LULC alternatives
Land-cover area summary
Detection summary
Per-object results
Analytical hierarchy
| Level | Question answered | Model / training domain | Output |
|---|---|---|---|
| Scene | What broad LULC type best characterizes this image? | ConvNeXT-Tiny / EuroSAT Sentinel-2 RGB | Ranked probabilities + entropy |
| Pixel | Which cover class is predicted at each pixel? | Mask2Former / OpenEarthMap | Overlay, mask, pixel shares |
| Object | Where are supported discrete objects? | YOLOv8n / NWPU VHR-10 | Boxes, counts, CSV, pixel GeoJSON |
Interpretation guardrails: EuroSAT is a European Sentinel-2 scene dataset; classification may shift on other sensors, regions, resolutions, or crops. Pixel shares are not automatically physical ground-area shares. Pixel-coordinate GeoJSON is not georeferenced. Models can miss small or obscured objects. Do not use outputs alone for legal, surveillance, emergency, navigation, or safety-critical decisions.