search

mlfpm / deepof

DeepLabCut based data analysis package including pose estimation and representation learning mediated behavior recognition
MIT License
39 stars 6 forks source link

Problem with animals' IDs when creating project #6

Closed Lucas97223 closed 1 year ago

Lucas97223 commented 1 year ago

I'm trying to use the create method like follows: script creation project I encounter this error, that I don't know how to solve. ERROR DEEPOF The animal IDs are red and white, the project is based on a multi animal DLC tracking dataset. When i try to remove body parts, I have this error message:

lucasmiranda42 commented 1 year ago

Dear Lucas, thank you once again for using the package!

I see the iterative imputation is still not converging. This could be the case if there are many occlusions in the videos, but should not be a problem down the line (you may end up with clusters corresponding to poor imputation that you'll need to filter afterward).

Regarding the error, I see you used "red" and "white" in lowercase when creating the project, but the error sugests the IDs in the actual file are uppercase ("Red", "White"). Could this be the problem?

In case it's not, could you provide a minimal example file (just the header and the first few lines would do) of the tracking files you're attempting to load? That would be of great aid to help you debug.

Best, and thank you once again! Lucas

Lucas97223 commented 1 year ago

I still have the same problem. Here is a link to one of the H5 files that I use coming from the tracking processing. https://drive.google.com/drive/folders/1HfxRxyvKVDq2iDae90UNeZPGKgM6_gA_?usp=sharing

lucasmiranda42 commented 1 year ago

Hi Lucas,

Thank you very much for uploading the file!

Upon inspecting it, it seems that your body parts are not labelled (i.e. they are called 'b1' to 'b8'). DeepOF fails because it has unfortunately no way to know what these tagged body parts are. Furthermore, at the moment, the package expects a given set of labels to fully function. Please refer to the landing page of the documentation (under 'What you need') for details, and feel free to keep asking questions, of course!

Best, Lucas