Hi @JulianAlvarezdeGiorgi, thanks a lot for your PR, I'm reviewing it! The UI part looks good so far.
Could you just clarify what your adapt_labeled_data_to_new_project function does?
Unless something has changed, body parts that are newly added to the config are seamlessly handled (and old data are converted to the newer multiindex format under the hood with guarantee_multiindex_rows).

Hello @jeylau , I created this function because I had originally created a dataset with 14 body parts. Later I decided to make another project, where I kept the body parts of the previous project (these 14) but added new body parts. As I already had many frames with the 14 labels, I wanted to be able to transfer them to the new project, and only have to label the new body parts. For this, I created a new project, with all the body parts I was interested in (old and new), I kept the “labeled-data” folder of the old project, and I used the adapt_labeled_data_to_new_project function to add the new body parts to the .csv and .h5 files, so I could label the new bodyparts from the GUI.

I didn't manage to do this by adding the new body parts to the config file, but I may have skipped a step.

I understand better. Normally, after adding new body parts to your config and opening the labeling GUI, you should automatically be warned that new body parts were detected, and asked whether or not you want to display all of them or only the new ones while labeling.

Is this true with DLC 2.3.9?
Because for me, it was not like this, if I extract new frames I can label all the body parts, but on the already extracted and labeled frames, I don't have the option to label the new body parts.

Yes, it's a feature we had even before the napari-deeplabcut labeling plugin. I tested it this morning again. Maybe try pip install -U napari-deeplabcut just to make sure you're running the latest version.

If useful, seems okay to me!

Perfect, I'll try updating deeplabcut then. I've just run the black formatting, however, it changed a lot of files that I didn't even touch, should I have run it with a specific flag/configuration?

Thank you very much for your help and your willingness to collaborate !