Can you provide the test set about Visa, ELPV, SDD, AITEX? about inctrl

[Diana1026]

Hello, the normal samples are randomly selected from the training set of the target dataset. Hence, you can just use the whole test set and do not need a data split. Please refer to the content in Sec 4.1 of my paper. Thank you.

[lalisa2020]

Thanks for your reply!

However, I find that several datasets do not provide standard training and testing sets, requiring users to split them themselves.

For example, the original headCT dataset consists of 100 normal slices and 100 abnormal slices, but in Table 5, your article uses the headCT dataset with 25 normal and 100 abnormal slices in the test set. The original SDD dataset contains 347 normal and 52 abnormal slices, but in the test set you used 286 normal and 54 abnormal slices.

[Diana1026]

You should use the convert scripts to convert the datasets to MVTec AD format. Please follow the guidance in README.