RuntimeError: stack expects each tensor to be equal size, but got [1, 64, 64, 64] at entry 0 and [4, 64, 64, 64] at entry 3

[ankitabuntolia]

I am following TorchIO Tutorial for my 3D OCT Image Segmentation task.

Background: All of my images and labels are of size 512x496x49 In the label I have 4 classes (including a background as 0 and all others are 1,2,3)

I am facing error of size mismatch of tensor.

Can someone tell me where am I doing wrong?

RuntimeError                              Traceback (most recent call last)
[<ipython-input-64-5b4812ec9e10>](https://localhost:8080/#) in <module>()
----> 1 one_batch = next(iter(training_loader))

3 frames
[/usr/local/lib/python3.7/dist-packages/torch/_utils.py](https://localhost:8080/#) in reraise(self)
    432             # instantiate since we don't know how to
    433             raise RuntimeError(msg) from None
--> 434         raise exception
    435 
    436 

RuntimeError: Caught RuntimeError in DataLoader worker process 0.
Original Traceback (most recent call last):
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/worker.py", line 287, in _worker_loop
    data = fetcher.fetch(index)
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/fetch.py", line 52, in fetch
    return self.collate_fn(data)
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/collate.py", line 74, in default_collate
    return {key: default_collate([d[key] for d in batch]) for key in elem}
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/collate.py", line 74, in <dictcomp>
    return {key: default_collate([d[key] for d in batch]) for key in elem}
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/collate.py", line 74, in default_collate
    return {key: default_collate([d[key] for d in batch]) for key in elem}
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/collate.py", line 74, in <dictcomp>
    return {key: default_collate([d[key] for d in batch]) for key in elem}
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/collate.py", line 56, in default_collate
    return torch.stack(batch, 0, out=out)
RuntimeError: stack expects each tensor to be equal size, but got [1, 64, 64, 64] at entry 0 and [4, 64, 64, 64] at entry 3

My Code

training_transform = tio.Compose([
    tio.ToCanonical(),
    # tio.Resample(4),
    tio.CropOrPad((64, 64, 64)), # tuple: (W, H, D)
    # tio.RandomMotion(p=0.2),
    # tio.HistogramStandardization({'image': landmarks}),
    # tio.RandomBiasField(p=0.3),
    # tio.ZNormalization(masking_method=tio.ZNormalization.mean),
    # # tio.RandomNoise(p=0.5),
    # tio.RandomFlip(),
    # tio.OneOf({
    #     tio.RandomAffine(): 0.8,
        # tio.RandomElasticDeformation(max_displacement=2): 0.2,
    # }),
    tio.OneHot(),
])

validation_transform = tio.Compose([
    tio.ToCanonical(),
    # tio.Resample(4),
    tio.CropOrPad((64, 64, 64)),
    # tio.HistogramStandardization({'image': landmarks}),
    # tio.ZNormalization(masking_method=tio.ZNormalization.mean),
    tio.OneHot(),
])

num_subjects = len(subjects_dataset)
num_training_subjects = int(training_split_ratio * num_subjects)
num_validation_subjects = num_subjects - num_training_subjects

num_split_subjects = num_training_subjects, num_validation_subjects
training_subjects, validation_subjects = torch.utils.data.random_split(subjects_list, num_split_subjects)

training_set = tio.SubjectsDataset(
    training_subjects, transform=training_transform)

validation_set = tio.SubjectsDataset(
    validation_subjects, transform=validation_transform)

print('Training set:', len(training_set), 'subjects')
print('Validation set:', len(validation_set), 'subjects')

Dataloader

training_batch_size = 16
validation_batch_size = 2 * training_batch_size

training_loader = torch.utils.data.DataLoader(
    training_set,
    batch_size=training_batch_size,
    shuffle=True,
    num_workers=num_workers,
)

validation_loader = torch.utils.data.DataLoader(
    validation_set,
    batch_size=validation_batch_size,
    num_workers=num_workers,
)

[fepegar]

Hi, @ankitabuntolia. Thanks for reporting this. Can you please create a bug report and ping @snipdome in it?

[ankitabuntolia]

Thank you, I have reported it at #848