Hi @fehmi.simsek,

I see that you are getting a lot of deprecation warnings because you are using a combination of eo-learn<1.0.0 and newer versions of its dependency packages, e.g. numpy>=1.20. There are 2 options what you can do about it:

• 
  

  Upgrade your code to use the latest eo-learn version, at the moment this is version 1.0.2. For this I recommend checking release notes and updated documentation examples to see what changed in the new versions. But once you do that your code will be compatible with the latest versions of all dependency packages.

  
  


• 
  

  Downgrade your eo-learn dependency packages until the warnings disappear. I think you’ll have to downgrade numpy, geopandas, shapely, and probably a few more.

  
  

Thank you Mr. Matej,

I will try it

Hi,

I updated eo-learn successfully

After update when i run notebook i saw this error

How can i fix ?

Hi,

the ExportToTiff task has been renamed to ExportToTiffTask. There were quite a few code-breaking changes in eo-learn 1.0.

Sadly, lcms is not compatible with eo-learn 1.0, however we have since released eo-grow (repo is here), the successor of lcms.

Hi @fehmi.simsek!

Good to hear from you and good to hear that you are still using eo-learn and our other related packages. To sum up what others have already said:

1. 
   

   You are using the trlcms package, which is a specific version of our lcms package. These packages are quite old at this point and were not kept up to date. Unfortunately this means that you are stuck with keeping all related packages frozen to the older versions for the code to work.

   
   


2. 
   

   Another option, mentioned by @maleksandrov, is to update all the packages, but this means you have to deal with solving tedious issues like the problem above, which requires getting your hands dirty with the code in order to make it work.

   
   


3. 
   

   As @ziga.luksic has mentioned, our old lcms package has been converted to a new, publicly available eo-grow package, which has the same goal - using pipelines in order to process EOPatches from an area of interest. A jump from trlcms to eo-grow would require doing changes in the configs and perhaps some pipelines would need to be ported from tlcms to eogrow. This package also has better documentation and examples.

   
   

I believe the last point is the most stable and straightforward, however it requires some work. You mentioned you have problems with downloading the eopatches. Perhaps you can look at the workshop linked below and decide for yourself if this is worth your time. If you run into any issues as a user you can create posts on this forum or create tickets on the eo-grow GitHub repo, however, we cannot help much if you decide to go with either of the remaining steps, which would require finding correct versions of the used packages.

eo-grow/examples/workshop at main · sentinel-hub/eo-grow

main/examples/workshop

Earth observation framework for scaled-up processing in Python

Regarding the specifics of your issue above: The code produces CLM and CLP cloud masks, but in the meantime we started to provide CLM and CLP similarly as bands, from the service, which you can just download. It can be provided via the additional_data parameter here. You can also read more here:

Cloud masks were the most common requests we’ve been getting for the last few years. We provided many options - Hollstein’s custom script, Braaten-Cohen-Yang cloud detector, L2A scene classification, open-source s2cloudless library. However, none of these was good and simple enough.
This changes now as we have processed the complete archive with our s2cloudless.
You can now simply use “bands” CLM (cloud mask) or CLP (cloud probablity).
More information in the documentation:


Some more back…

I know this is a lot of information, so take your time in order to make a decision. Feel free to ask questions if you get stuck.

Best of luck!

Cheers,

Matic