← Marketplace

Field detection in Europe


Coming back soon: currently not available. An algorithm that detects agricultural field boundaries in Central Europe from SPOT imagery. The algorithm automatically delineates agricultural fields visible in a satellite image, producing a layer of polygons to describe them. Field detection can be used for environmental monitoring, disaster response, and land cover classification.

Technical information

SourceSPOT tasking or catalog collections
Compatibility- The STAC item should be a SPOT image of the display radiometric processing level
- The STAC item should have been added to storage in 2023 or later
Required parameters- Console: If your chosen input data meets the above specifications you will be able to run your job
- API: You will need to input an Output title and the Input data
Output data formatGeoJSON file: with the extracted field boundaries as a MultiPolygon
OutputThe result will be added to your account as a STAC item in a new STAC collection. You can retrieve the resulting data in one of the following ways:
- Open the console, go to Data managementJobs
- Retrieve the results via the API

Algorithm performance and training data

The model has been developed using agricultural environments typical for central Europe and in particular the following European territories:

  • Austria
  • Belgium
  • Czechia
  • Denmark
  • France
  • Germany
  • Ireland
  • Luxembourg
  • United Kingdom

For these territories, the Europe model has achieved a mean field F1 score of 0.85. Users can apply the Europe model to other territories with similar agriculture environments but Airbus cannot guarantee the algorithm has the same level of accuracy if applied to other territories. For Iowa, please use the Field detection in Iowa algorithm.

The multi-territory Field detection algorithm has been developed to apply a single model to a selection of central European countries by using a variety of high-resolution satellite images to ensure a high level of accuracy and to account for different locations that have distinct features. A number of different scenarios were represented in the training data, including: different points in the growing season, all possible terrain types, all possible features that could be encountered (including those unrelated to agriculture), and the use of data augmentation. This was to ensure the training data was of a good quality, so the algorithm produced results with high precision and recall when applied to SPOT imagery, and to reduce the number of false positives by including the identification of non-agricultural features.

Terms & conditions

View the End User License Agreement conditions

Ready to get started?

Explore UP42 now.

Sign in or create an account