Hello,
I'd like to express my gratitude for this incredibly valuable library and commend you for your exceptional work.
I'm curious to know if there's a way to manually designate the training, validation, and test sets as input.
While I've explored the Folder and FolderDataset classes, it seems that they don't fulfill this requirement, as they perform their own data split based on predefined parameters and a seed.
Therefore, I'm wondering if Anomalib features a mechanism that enables me to precisely specify the images for training, validation, and testing. Because I want to do some experiments by varying the images used for training and validation and test.
I'm looking for something equivalent to (if it exists):
Data(train_path =/path/to/train, validation_path=/path/to/validation, test_path=/path/to/test)
Thank you in advance.