🚀 Replace albumentations with torchvision transforms

[djdameln]

📝 Description

PR for replacing albumentations transforms with torchvision transforms

✨ Changes

• Use torchvision v2 transforms instead of albumentations transforms throughout the code base.
• Load images as torch tensors instead of numpy images in dataset __getitem__ methods.
• Use tv_tensors Mask format for GT masks
• Transforms are included in the exported model. This is done by wrapping the transform and the torch model in the new InferenceModel class before exporting.
• The datamodules take a Transform object as input to directly set the transform that is applied to the input images.
• The model implementer can define a default model-specific transform using the configure_transforms method of the Lit model. When no transform is passed to the datamodule, this default transform will be used. The default model transform may use the image_size parameter from the datamodule to resize the images to the size specified by the user.

To be addressed:

• Inferencers still use albumentations
• Current design does not support model-specific transforms (i.e. defining a default transform from the lightning model).
• Still need to implement the model-specific default transforms for most models. This can be done later as they now just use the default one from the base module.
• Docstrings in the data classes need to be updated to reflect the new signatures.
• Patchcore fails to export because of the CenterCrop transform. We may have to implement our own version of this transform to allow exporting.
• Some models still have the input_size parameter, which needs to be removed.
• CLI tests are failing because the export API has changed.
• The ConfigAdapter needs to be updated to reflect the new datamodules API.

Select what type of change your PR is:

• 🐞 Bug fix (non-breaking change which fixes an issue)
• 🔨 Refactor (non-breaking change which refactors the code base)
• 🚀 New feature (non-breaking change which adds functionality)
• 💥 Breaking change (fix or feature that would cause existing functionality to not work as expected)
• 📚 Documentation update

✅ Checklist

Before you submit your pull request, please make sure you have completed the following steps:

• 📋 I have summarized my changes in the CHANGELOG and followed the guidelines for my type of change (skip for minor changes, documentation updates, and test enhancements).
• 📚 I have made the necessary updates to the documentation (if applicable).
• 🧪 I have written tests that support my changes and prove that my fix is effective or my feature works (if applicable).

[samet-akcay]

Thanks for creating this pr. I've added my initial comments.. Will go for another round

[ashwinvaidya17]

I didn't run this yet but overall I like this over albumentations. It will be one less dependency.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[samet-akcay]

Looking great! Thanks a lot!