Land Classification of Ukraine

Hello, I need a photo with a classification from eo-learn of any small part of Ukraine. What is the easiest way I can get it? Give detailed instructions or a link to a current example, please. Unfortunately, it is very difficult for me to understand your library, so I will be glad if you help me.
I saw that there are custom scripts. Maybe they will help me, but I have no idea how to use them.
(CNES land cover classification visualisation script | Sentinel Hub custom scripts)
how to implement it in code?

Hi Andrew,

Thanks for the question, I would recommend going through the example workflow on the EO Learn Documentation site first. The land use/ land cover example should be particularly relevant to your use case. This will help you understand the EO-Learn library and it should be fairly simple to apply this example to your own area of interest in Ukraine.

If you have some more specific questions after that, we’ll be happy to answer!

I got acquainted with this example, but since I am not very strong in this, it is difficult for me to understand how to manipulate this file. I would be very grateful for detailed step-by-step instructions on how to substitute the data of Ukraine into the trained model, in what format they should be, and where to find them.

Hi Andrew,

The example that I shared shows you how to train your own model, using land cover reference data from Slovenia. The format of the data is in Geopackage but this could be substituted for another OGR compliant vector format like geojson or a shapefile.

Therefore, there are two steps you need to make to substitute your AOI to the example:

  1. Replace the vector of Slovenia with your area of interest.
  2. Find a reference dataset to train your model with the relevant land cover classes for your use case.

As for finding reference data that you could use to train a model for Ukraine, this is not something that I have much knowledge about and you are probably better placed to conduct this research.

Hope that this information helps you out :slight_smile: