Are masks needed if we use classification models like dfm ?

[MhdKAT]

I tried to run dfm to classify my images as anomalous/non anomalous. I made my dataset in MVTec hierarchy but I get an error when i run the train script with the error being that the trainer can't access segmentation masks. Are they mandatory when the task is to classify? Below is a the error I get.
File "C:\Users\hp\anomalib\env\lib\site-packages\torch\utils\data_utils\fetch.py", line 44, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "c:\users\hp\anomalib\anomalib\data\mvtec.py", line 256, in getitem
mask = cv2.imread(mask_path, flags=0) / 255.0
TypeError: unsupported operand type(s) for /: 'NoneType' and 'float'

[LukasBommes]

I just got the same error when trying to run task: classification on an mvtec-like dataset (format: mvtec) without masks provided.

In my opinion, it would make most sense to fix the folder dataset to correctly deal with custom datasets that already have a train/test split (this currently does not work, see here). In that case, there would be no need to use format: mvtec for a custom dataset without masks.

[samet-akcay]

@MhdKAT, it's the same issue that was found in folder dataset. Need to propose the same fixes to address the issue.
@LukasBommes, yes that's on our todo list, but didn't have a chance to work on it this week unfortunately.

[MhdKAT]

I see ! Thanks for the clarification @LukasBommes, @samet-akcay.