Unexpected key(s) in state_dict: "pixel_metrics.AUPRO.fpr_limit".

[1094558673]

Hello, I used pixel's AUPRO metric in the training process. When I run inference code "python tools/inference/lightning_inference. py -- -- config src/anomalib/models/efficient_ad/config.yaml --weights ...” Unexpected key(s) in state_dict: "pixel_metrics.AUPRO.fpr_limit"." Error.

I saw this in an issue a few months ago, but it doesn't seem to have been fixed in the master branch.

[blaz-r]

Seems from #1055 that it's not merged as it will soon removed from module.
Any update on this @ashwinvaidya17 ?

[ashwinvaidya17]

Our inference entrypoints will change a bit with Anomalib 1.0 version. Addressing this is also a part of the refactor.

[bluehyena]

I've gone through several training sessions, but I'm facing challenges with testing on this particular issue. Could you please advise on how to proceed?

[blaz-r]

Hello @bluehyena, check #1055 for a temporary fix.

[samet-akcay]

If this PR still up-to-date, I think we should merge this. I thought we had merged this, but looks like we forgot :)

[blaz-r]

@samet-akcay I believe this issue can be closed as it was addressed in #1055.