AttributeError: 'MVTecDataset' object has no attribute 'transform_std'

[William9Baker]

What's wrong with this error?

[zsffuture]

I also encountered the same problem, did you solve it?

[zsffuture]

I solved this problem, because backbone is based on imagenet dataset training, so the mean and variance is imagenet, you can add it in src\patchcore\datasets\mvtec.py file can be seen at the top, and then modify it to add it.

[PauloFavero]

I fixed this issue in the simplenet repo with the following pull request: DonaldRR/SimpleNet#20
To fix the issue here in patchcore just follow the modifications in simplenet dateset folder.

[romanvelichkin]

Long story short:

in patchcore/datasets/mvtec.py in class MVTecDataset in def __init__ add this rows:

self.transform_std = IMAGENET_STD
self.transform_mean = IMAGENET_MEAN

[IESSTTJP]

I am using a different dataset, but adding those two sentences still reports the same error, can you please tell me in which rows they are added?

[DanielWell]

I am using a different dataset, but adding those two sentences still reports the same error, can you please tell me in which rows they are added?

i also try this, but also report the same error, as you are. Maybe when run_patchcore.py->image_trainsform function to run , but mvtec.py->mvtec not init. Now, in run_patchcore.py->image_transform， this function,
def image_transform(image):
IMAGENET_MEAN = [0.485, 0.456, 0.406]
IMAGENET_STD = [0.229, 0.224, 0.225]
# in_std = np.array(
# dataloaders["testing"].dataset.transform_std
# ).reshape(-1, 1, 1)
# in_mean = np.array(
# #dataloaders["testing"].dataset.transform_mean
# ).reshape(-1, 1, 1)
in_std = np.array(IMAGENET_STD).reshape(-1, 1, 1)
in_mean = np.array(IMAGENET_MEAN).reshape(-1, 1, 1)
image = dataloaders["testing"].dataset.transform_img(image)
return np.clip(
(image.numpy() * in_std + in_mean) * 255, 0, 255
).astype(np.uint8)