feat(XAI): Extending AUCMEDI XAI to cope with 3D data and batched inputs.

[Deathlymad]

• [x] GradCAM
• [x] GradCAM++
• [x] Guided Backpropagation
• [x] Guided GradCAM
• [x] Integrated Gradients
• [x] Saliency
• [x] LIME
• [x] Occlusion Sensitivity

[Deathlymad]

Guided GradCAM and the Decoder can probably be measurably accelerated by parallelizing the Resize operation or making it batch compatible.

[muellerdo]

Hey @Deathlymad,

please be aware that your git commits do not follow our git commit conventions.

Check out here: https://frankkramer-lab.github.io/aucmedi/getstarted/contribution/#special-contribution-notes-for-aucmedi and here: https://www.conventionalcommits.org/en/v1.0.0/

[Deathlymad]

aaaaaaaaaaaaaaaaaaaaaaaaaa

[Deathlymad]

This was intended to update the commit messages. It appearsr that git misinterpreted the operation.

[Deathlymad]

Remaining Errors related to AutoML. Please advice on how to proceed.

[Deathlymad]

commited fix to this branch that should resolve issues related to pandas 2.0 update

[Deathlymad]

The coverage error is linked to a missing token

[muellerdo]

can you reset the last pandas commit for the evaluation on focusing on just the xai decoder? :)

avoiding large PRs changing various aspects and focusing on a one PR:one-issue-fix policy.

-> checkout #209

[Deathlymad]

@muellerdo reverted the pandas fix. I believe that my edits are limited to the XAI files.

[muellerdo]

Stale, as batch-wise XAI decoder & methods revealed functionality issues. Reopen if working solutions for these issues are found.