AI4EO: AI for Remote Sensing and Earth-Observation Data

AI4EO is the Earth-observation service line of Quadratic, a geomatics consultancy based in Voroux-lez-Liers (Liège province), Belgium. It designs, develops and maintains custom geoprocessing routines and machine-learning workflows that turn satellite imagery into geodata and statistics for a client's own monitoring or planning needs, delivered either as a prototype or as a production system.

Quadratic lists Sentinel-1, Sentinel-2, Landsat and Pléiades among the satellite data it works with. Its toolset is mainly open source: Orfeo ToolBox, SNAP, GRASS GIS and QGIS for image processing and analysis, PostgreSQL/PostGIS for spatial storage and Python for automation, with machine learning applied to classification and clustering.

The company's published references describe three projects of this kind, without naming the clients:

• For an environmental research institute, routines that process Sentinel-2 image time series and produce geo-statistics for the systematic monitoring of around 2,000 areas of interest.
• For an agro-industrial grain collector, analysis of Sentinel imagery with machine learning to identify the main crop on each agricultural plot (soft wheat, barley, rapeseed, potatoes, beets). The client uses these geodata to plan harvest logistics, support sales and decide where to build new facilities.
• For a smart-grid partner, research into using Sentinel-1 radar and Sentinel-2 optical time series with automatic change detection to estimate vegetation growth near high-voltage power lines.