Fix to torch_inferencer pre_process trying to use 'to(device)' on a n…

[Sbaffinator98]

Fix to torch_inferencer pre_process trying to use 'to(device)' on a n…umpy istance with broken shape

📝 Description

• With 1.0.0 the torch inferencer is broken, when working on 501a for example, if we convert the export and inferencer of the model to the TorchInferencer we will get broken result caused by doing numpy.to(device) in the pre_process of the inferencer. Being .to(device) a tensor method it is not possible to be working.
  Provided the fix by making a .transpose of the ndarray coming in and exporting the result with torch.from_numpy(ndarray).to(device) instead of ndarray.to(device)

On my fork on branch 501a_torch_fix can be found the .ipynb with the implementation on the cubes training/inference with OpenVino but also with the TorchInferencer

✨ Changes

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
• 🔒 Security 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).

[Sbaffinator98]

Thie is my first time asking for a pr on any open source project, i am open to correct and explain what i did. Thank you

[samet-akcay]

Thie is my first time asking for a pr on any open source project, i am open to correct and explain what i did. Thank you

Welcome to the community, @Sbaffinator98! Thanks a lot for your contribution!

I'll be running the tests now, and we will be reviewing your PR shortly. Thanks again!

[samet-akcay]

Hi @Sbaffinator98, thank you for addressing this. I've added some suggestions to fix the code-quality checks. If you want to re-produce it on your side, you could do the following from your terminal:

pip install pre-commit
pre-commit run -a

[Sbaffinator98]

Hello i saw that you guys commited some clean-code. I see that is still failing the checks. I dont see any torch_inferencer related issue. What should i do further?

[samet-akcay]

Hello i saw that you guys commited some clean-code. I see that is still failing the checks. I dont see any torch_inferencer related issue. What should i do further?

The tests are failing for some reason here
https://github.com/openvinotoolkit/anomalib/actions/runs/8367192205/job/22909054968?pr=1816

I'll check why they fail

[samet-akcay]

Hi @Sbaffinator98, thanks for your patience, and apologies for going over this pr a bit late.
With the recent v2 update, metadata is no longer required in torch inference. I'm therefore closing this PR