Adding RandCoarseDropoutd causes a crash

[dzenanz]

If I add RandCoarseDropoutd to my list of transforms, the program encounters ValueError: Sequence must have length 2, got 3. The "image" has a shape (7, 368, 640) and "label" has a shape (1, 368, 640). What am I doing wrong?

    train_transforms = Compose(
        [
            LoadImaged(keys=["image", "label"]),
            modelio.ExtractChannelTransform(keys=["image"], channel_ids=channel_ids),
            AddChanneld(keys=["label"]),
            RandFlipd(keys=["image", "label"], prob=0.5, spatial_axis=1),
            RandFlipd(keys=["image", "label"], prob=0.5, spatial_axis=0),
            RandCoarseDropoutd(
                keys=["image", "label"],
                fill_value=0,
                prob=0.5,
                # sizes are for channel, height, width
                spatial_size=(-1, 32, 32),
                max_spatial_size=(-1, 64, 128),
                holes=4,
                max_holes=16,
            ),
            RandAffined(
                keys=["image", "label"],
                prob=0.75,
                padding_mode="zeros",
                rotate_range=np.pi,
                translate_range=15,
                shear_range=0.25,
                scale_range=0.25,
                mode=("bilinear", "nearest"),
            ),
            ToTensord(keys=["image", "label"]),
        ]
    )

and trace:

Traceback (most recent call last):
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\torch\utils\data\dataloader.py", line 530, in __next__
    data = self._next_data()
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\torch\utils\data\dataloader.py", line 1224, in _next_data
    return self._process_data(data)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\torch\utils\data\dataloader.py", line 1250, in _process_data
    data.reraise()
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\torch\_utils.py", line 457, in reraise
    raise exception
RuntimeError: Caught RuntimeError in DataLoader worker process 0.
Original Traceback (most recent call last):
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\transforms\transform.py", line 92, in apply_transform
    return _apply_transform(transform, data, unpack_items)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\transforms\transform.py", line 60, in _apply_transform
    return transform(parameters)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\transforms\intensity\dictionary.py", line 1509, in __call__
    self.randomize(d[self.keys[0]].shape[1:])
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\transforms\intensity\dictionary.py", line 1504, in randomize
    self.dropper.randomize(img_size=img_size)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\transforms\intensity\array.py", line 1711, in randomize
    size = fall_back_tuple(self.spatial_size, img_size)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\utils\misc.py", line 179, in fall_back_tuple
    user = ensure_tuple_rep(user_provided, ndim)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\utils\misc.py", line 136, in ensure_tuple_rep
    raise ValueError(f"Sequence must have length {dim}, got {len(tup)}.")
ValueError: Sequence must have length 2, got 3.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\torch\utils\data\_utils\worker.py", line 287, in _worker_loop
    data = fetcher.fetch(index)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\torch\utils\data\_utils\fetch.py", line 49, in fetch
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\torch\utils\data\_utils\fetch.py", line 49, in <listcomp>
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\data\dataset.py", line 95, in __getitem__
    return self._transform(index)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\data\dataset.py", line 667, in _transform
    data = apply_transform(_transform, data)
  File "C:\Users\Dzenan\miniconda3\envs\deep_learning\lib\site-packages\monai\transforms\transform.py", line 116, in apply_transform
    raise RuntimeError(f"applying transform {transform}") from e
RuntimeError: applying transform <monai.transforms.intensity.dictionary.RandCoarseDropoutd object at 0x00000232CB7368B0>

[dzenanz]

The problem was in spatial sizes. I should not have supplied channel dimension. Instead of

                spatial_size=(-1, 32, 32),
                max_spatial_size=(-1, 64, 128),

this works (in the sense that it does not crash):

                spatial_size=(32, 32),
                max_spatial_size=(64, 128),