Geoalert platform
Published in

Geoalert platform

Mapping agricultural fields with AI!

If you have projects/solutions regarding crop analysis or Inventory of the croplands, you know that you need precise boundaries of each agricultural field for advanced analytics. And if a project coverage is large enough, you have to commit a lot of time and money to digitize high-resolution satellite imagery.

Let’s do some estimates:

  • Let’s take Ukraine region fields ~60.000​​​​ fields within the area of 12.208​​​​ sq. km
  • Usually, it takes approximately ~2m per one field for professionals to outline its boundaries. It depends on how curvy the field is.
  • It will take up to 2 000 hours of continuous labor, or ~12.5 months of the work of professional cartographer to digitize everything. That means that you need to do a lot of labor-intensive work just to start working on crops production analytics

But what if we told you that you could start working on analytics on the same day you get the area of interest of your project?

Let’s have a sneak peek at how to do it!

We are here to help you to map faster!

The significant part of that tedious mapping work could be done using our in-house AI model, and you need a fraction of that cartographers team to adopt the results produced by AI. And the awesome thing is that you can estimate if the mapping of the field boundaries within your area of interest can be assisted by AI right now. Just try the “Agriculture fields” model in

A scene from the sensational RUSSIAN CYBERPUNK FARM

We trained neural networks to delineate agricultural fields boundaries using high-resolution imagery from Maxar and medium-resolution imagery from Sentinel 2. The area of 60K sq. km could be processed by our platform in just 2 hours using Sentinel 2 with resolution of 10 [m/pixel], or in a single day using our high-resolution model with resolution 1.2 [m/pixel].

AI is in charge of extraction of approximate field boundaries at a great speed and at a large scale, and you need to contribute much less time to validate and apply these results to your project.

Quick demo of the new AI-model in

How to use “Agriculture fields detection”:


For initial purposes we deployed the model that performs better at the small fields with an area more than 2 ha and less than 200 ha.

Look it by yourself without creating an account at — Link to Demo processings:

AI-mapping of agriculture fields — sample area of Morocco

And then you can do your own processing as the following:

  1. Create a new flow
  2. Press + at the “select raster source” step
  3. Upload your geojson with coordinates of the polygon area or draw a rectangle on the map
  4. Select Agriculture fields AI model
  5. Run flow and make and make yourself a nice cup of tea
  6. Select “Open with” and enjoy!

Further work

Of course it’s only the starting point we are at and we’ll keep on experimenting with the new methods, adding more samples to make the new model more robust and accurate for different patterns of agriculture lands. Let us know what you think and what’s the possible application to your workflow or stay tuned for new updates!


  • — The Geoalert platform for AI-mapping
  • Previous post about agricultural fields boundaries delineation




About Geoalert platform for new applications of the Earth Observation data powered by Deep learning analysis

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store


We apply Machine learning to automated analysis over Earth observation data

More from Medium

AI-assisted mapping of crop fields using free Sentinel imagery

AI-based Super resolution and change detection to enforce Sentinel-2 systematic usage

Types of satellite networks used in IoT solutions

Detecting Agricultural Croplands from Sentinel-2 Satellite Imagery