How to perform data augmentations in custom datasets

[microchila]

What is the motivation for this task?

data enhancement in custom datasets

Describe the solution you'd like

Some effective data enhancement methods such as UDA can be performed

Additional context

NOPE

[abc-125]

You can try custom transforms. Although most anomaly detection papers do not use transforms at all because more robust models can be actually worse at detecting anomalies.

[zhongxiaj]

just like this

eval_transform = Compose(
    [
        v2.Resize((256, 256), antialias=True),
       v2.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
    ],
)    

datamodule = Folder(
    name= 'bottle',
    # root=dataset_root,
    normal_dir="good",
   
    abnormal_dir="test",
    
    normal_split_ratio=0,
    mask_dir=None,
    # mask_dir=None,
    train_batch_size = 1,
    # eval_batch_size=1,
    num_workers=8,
    train_transform = eval_transform,
    eval_transform = eval_transform,
   
    test_split_mode=TestSplitMode.NONE,
    
    task="segmentation",
 
)