Building and road extraction
Takes satellite or drone imagery and returns building polygons and road lines.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Service 02 of 06
GeoAI and Earth observation
Models that read satellite, drone and street imagery and turn pixels into counts, classes and change.
GeoAI is machine learning applied to maps and imagery. Our models read satellite, drone and street-level images and turn them into things a business can count: buildings, roads, fields, crops, solar panels, shops and change over time. They also estimate population and demand, and let people ask the map questions in plain language.
Every model is checked against field samples and hand-mapped areas before its output is used, and each one ships with a short model card that says what data it learned from, where it was tested and where it is weak.
What's inside · 10 offerings
Each one built to answer a question someone has to decide.
Land use and land cover classification
Takes sentinel-2 time series and returns class per 10 m pixel.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Field boundaries and crop type
Takes sentinel-1 and Sentinel-2 time series, field labels and returns field polygons with crop type.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Change detection
Takes image time series and returns areas and dates of change: new buildings, forest loss, cleared land.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Object detection and counting
Takes high-resolution, drone or street-level imagery and returns points for solar panels, generators, vehicles, shops and signage.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Population and demand estimation
Takes building footprints and use, census totals, wealth proxies and returns people and demand per grid cell.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Address parsing and geocoding
Takes written addresses and landmark directions and returns coordinates and the matched building or place.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Ask the map
Takes a question in plain language and returns an answer with the map, the query used and the layers behind it.
- How it is delivered
- Delivered as layers, through the API or inside a product.
Farm boundary and crop monitoring
“What is planted on each field, and how is it doing this season?”
Maps field boundaries, crop type and crop condition through the season from satellite imagery and field samples.
- How it is delivered
- Web app and mobile field app. Subscription per hectare band or season.
- A first project
- Boundaries and condition for one scheme's fields
Disaster and flood response mapping
“Who is affected right now, and which roads are still open?”
Produces flood extent from radar within days of an event and overlays population, buildings and roads.
- How it is delivered
- Rapid maps and web viewer. Retainer or per activation.
- A first project
- Rapid flood extent product for one river basin
Start with one question.
Tell us the decision you need to make and where. We will come back with the smallest piece of work that answers it.
Other services
Location intelligence platforms
Your customers, assets, competitors and risks on one live map your teams actually use.
Spatial data products and layers
Ready-made and custom layers for demand, access, risk and infrastructure, delivered as files, tiles or API.
Spatial analytics and decision dashboards
Catchments, travel time, demand forecasts and optimisation, on dashboards that update as data arrives.
Field data and enumeration systems
Offline collection, address and asset registers, and census frames, checked on the ground.
Geo data engineering and integration
Geocoding, address matching, spatial databases and pipelines that plug location into your systems.