Mapflow

Mapflow is an AI mapping platform from Geoalert that extracts real-world objects from satellite, aerial and drone imagery and returns them as editable vector layers. It replaces manual digitising of buildings, roads and vegetation, and lets teams refresh only the parts of a map that have changed instead of buying a whole new dataset.

A user draws or uploads an area of interest, picks an imagery source and runs a model. Built-in sources are the Mapbox Satellite basemap (mainly Maxar imagery at about 0.5 m), ArcGIS World Imagery and a Global Mosaic built from Jilin satellite imagery; paid plans add search and ordering of commercial satellite imagery, and users can upload their own GeoTIFFs. Default models map building footprints (with type and height options), forest and tree cover with heights, and roads. Custom models cover construction sites, solar panels, agricultural field boundaries and other classes on request. Results download as GeoJSON or open directly in QGIS.

Mapflow is offered as a web app, an open-source QGIS plugin (199,000+ downloads), a REST API and a beta chat agent. It runs hosted by Geoalert, on separate AWS instances or on-premise. Enterprise services include custom model training, an SLA, powerline vegetation analysis (Geoalert states an accuracy of F1 0.6–0.8 depending on the vegetation class) and telecom map updates. Customers quoted on the site include IGECEM (the State of Mexico cadastre institute), Kontur, the Réunion planning agency AGORAH, India's National Institute of Urban Management, LOP, Island Green Power and ZorroEngine, a China-based 3D engine team that maps buildings and vegetation around airports. Plans run from a free tier through pay-as-you-go credits to monthly subscriptions.