[Bug]: Wrong normalization used for some models.

[blaz-r]

Describe the bug

Some of the models have wrong normalization setup in current version of code.
This list includes: DSR, DRAEM, EfficientAD, and rkde.

All of the above methods shouldn't use imagenet normalization, but they have the default method setup (which uses imagenet normaliztion):

anomalib/src/anomalib/models/components/base/anomaly_module.py

Lines 224 to 241 in 9482d34

	def configure_transforms(self, image_size: tuple[int, int] | None = None) -> Transform:
	"""Default transforms.
	The default transform is resize to 256x256 and normalize to ImageNet stats. Individual models can override
	this method to provide custom transforms.
	"""
	logger.warning(
	"No implementation of `configure_transforms` was provided in the Lightning model. Using default "
	"transforms from the base class. This may not be suitable for your use case. Please override "
	"`configure_transforms` in your model.",
	)
	image_size = image_size or (256, 256)
	return Compose(
	[
	Resize(image_size, antialias=True),
	Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
	],
	)

or in case of the EfficientAD it is manually set to imagenet:

anomalib/src/anomalib/models/image/efficient_ad/lightning_model.py

Lines 311 to 319 in 9482d34

	def configure_transforms(self, image_size: tuple[int, int] | None = None) -> Transform:
	"""Default transform for Padim."""
	image_size = image_size or (256, 256)
	return Compose(
	[
	Resize(image_size, antialias=True),
	Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
	],
	)

Dataset

Other (please specify in the text field below)

Model

Other (please specify in the field below)

Steps to reproduce the behavior

/

OS information

/

Expected behavior

All of the above (and potentially some other models, but I think those are the only four) should have normalization set to just get all the images to [0, 1] range.

Screenshots

No response

Pip/GitHub

GitHub

What version/branch did you use?

main

Configuration YAML

/

Logs

/

Code of Conduct

• I agree to follow this project's Code of Conduct

[youngquan]

Does using imagenet normalization affect the results much? Did you do any experiments? When I used draem, the anomaly maps rendered poorly. The anomalous regions are not rendered red. I'm still trying to figure out why.

[blaz-r]

Draem must not use imagenet normalization because the synthetic anomalies are in range 0-1, therefore images need to be normalized to 0-1 as well.