Geospatial Lab
Land-use classification, change detection, terrain segmentation, and wildfire risk via API. NASA Prithvi-EO + Clay Foundation + TorchGeo.
What's inside
Land-Use Classification
Classify land-use from satellite imagery using NASA Prithvi-EO, Clay Foundation, SatCLIP, or TorchGeo. Urban, forest, cropland, water, wetland, and more. Pretrained on 40 years of Earth observation data.
Change Detection
Compare two satellite images to detect deforestation, urbanization, flooding, or agricultural change. Returns change hotspots with confidence scores and magnitude.
Terrain Segmentation
Segment terrain into water bodies, vegetation, bare ground, and built-up areas. Returns mean elevation and max slope. Perfect for mining, forestry, and urban planning.
Wildfire Risk Prediction
Predict wildfire risk for any lat/lon with contributing factors (dryness, wind, vegetation moisture, temperature). Horizons from 1 hour to 30 days.
FAQ
What's Prithvi-EO?›
NASA/IBM's open-source geospatial foundation model, released on HuggingFace. Trained on 40+ years of Landsat and Sentinel imagery. Reconstructed global surface temperatures from just 5% of input data in benchmarks.
What's Clay Foundation?›
An open-source Earth observation foundation model from the Clay team. Designed as a community project with multi-sensor pretraining. Pairs well with Prithvi for ensembling.
Can I upload my own satellite images?›
Yes. Pass base64-encoded imagery to any endpoint. Supports RGB, multispectral, and standard satellite formats. Max 50 MB per image.
What's the Copernicus angle?›
The Copernicus Sentinel program generates petabytes of free satellite data annually. SciRouter gives you the inference layer — fetch imagery from Copernicus, process through our API, get land-use classifications in seconds.
Who uses this?›
Environmental researchers, urban planners, disaster response teams, mining companies, forestry services, and agtech platforms. Geospatial AI is projected at $8B+ by 2030 (32% CAGR).