🔨 Pass pre-trained from config to ModelLightning

[samet-akcay]

Description

• Following FIX: PaDiM didn't use config.model.pre_trained. #514, this PR adds pre_trained parameter to ModelLightning implementation.

Changes

• 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)
• This change requires a documentation update

Checklist

• My code follows the pre-commit style and check guidelines of this project.
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing tests pass locally with my changes

[djdameln]

I'm having some doubts about adding this parameter. For feature extraction based models such as padim and dfm, what would be the use case of having a non pre-trained backbone? Since the training does not involve finetuning the backbone weights, the normality model would be built upon random layer activations.

[samet-akcay]

I'm having some doubts about adding this parameter. For feature extraction based models such as padim and dfm, what would be the use case of having a non pre-trained backbone? Since the training does not involve finetuning the backbone weights, the normality model would be built upon random layer activations.

I don't know if there is a specific use-case for this. config.yaml file contains pre_trained flag. If we dont want this, maybe we could remove this from the config file itself.

Perhaps you could ask this question in PR 514 to confirm whether this actually has any use-case?

[djdameln]

Based on the discussion in the other PR (#514) I think this is a viable use case, so I'm happy to merge this one.

[ashwinvaidya17]

Thanks!