MIC-DKFZ / batchgenerators

A framework for data augmentation for 2D and 3D image classification and segmentation
Apache License 2.0
1.09k stars 221 forks source link

Publish new version of batchgenerator for using Misalignment DA #122

Open alexandreroutier opened 3 months ago

alexandreroutier commented 3 months ago

Hello,

I tried to install nnUNet flavour of https://github.com/MIC-DKFZ/misalignment_DA/ and ran some inference with it.

I had the following issue:

  File "<path>/nnunetv2/training/nnUNetTrainer/nnUNetTrainer_Misalign.py", line 20, in <module>
    from batchgenerators.transforms.spatial_transforms import SpatialTransform, MirrorTransform, MisalignTransform

but it turns out that I had the same issue with @ndebs: https://github.com/MIC-DKFZ/misalignment_DA/issues/1

However, it seems that this new transform is not yet published on pypi:

commit f53025bd05fe9642133ac539a0df3559d82415b4 (HEAD -> master, origin/master, origin/HEAD)
Date:   Tue May 28 10:16:34 2024 +0200
Info: use pandas for determining unique segmentation values, only compute unique values when needed
[...]
commit cb0c65300e6fc5f94a42e3e279e4577b9447b3d5
Date: Mon Dec 4 14:04:14 2023 +0100
Info: Adding Misalignment DA
[...]
commit 6859efd8cd59000896c0bcb6313e2b8e12bbb031 (tag: v0.25): bump version

If I am correct, would it be possible to publish a new version on Pypi instead of having to execute pip install --upgrade --force-reinstall git+https://github.com/MIC-DKFZ/batchgenerators.git after the installation of nnUNet?

Thanks in advance for your help.

Best, Alexandre