Stack expects each tensor to be equal size

1

Hello I got this error:
my code is

# 6. Prepare dataset

class dataset(Dataset):
    
    def __init__(self, data_dir, image_fns):
        self.data_dir = data_dir
        self.image_fns = image_fns
        
    def __len__(self):
        return len(self.image_fns)
    
    def __getitem__(self, index):
        image_fn = self.image_fns[index]
        image_fp = os.path.join(self.data_dir, image_fn)
        image = Image.open(image_fp).convert('RGB')
        image = self.transform(image)
        text = image_fn.split(".")[0]
        return image, text
    
    def transform(self, image):
        
        transform_ops = transforms.Compose([
            transforms.ToTensor(),
            transforms.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))
        ])
        return transform_ops(image)

cpu_count = mp.cpu_count()
print(cpu_count)

trainset = dataset(data_path, image_fns_train) 
testset = dataset(data_path, image_fns_test)
train_loader = DataLoader(trainset, batch_size=batch_size, num_workers=cpu_count, shuffle=True)
test_loader = DataLoader(testset, batch_size=batch_size, num_workers=cpu_count, shuffle=False)
print(len(train_loader), len(test_loader))


image_batch, text_batch = iter(train_loader).next()
print(image_batch.size(), text_batch)

And I got this error

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-67-521157e8a64b> in <module>()
      1 
----> 2 image_batch, text_batch = iter(train_loader).next()
      3 print(image_batch.size(), text_batch)

3 frames
/usr/local/lib/python3.7/dist-packages/torch/_utils.py in reraise(self)
    423             # have message field
    424             raise self.exc_type(message=msg)
--> 425         raise self.exc_type(msg)
    426 
    427 

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 47, in fetch
    return self.collate_fn(data)
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/collate.py", line 84, in default_collate
    return [default_collate(samples) for samples in transposed]
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/collate.py", line 84, in <listcomp>
    return [default_collate(samples) for samples in transposed]
  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 [3, 25, 20] at entry 0 and [3, 27, 23] at entry 1

Please help me with this, thank you so much

2

So each worker will essentially call your __getitem__ method batch_size number of times. Each time it’s getting an individual tensor. Afterwards, it will attempt to stack all of these tensors together to provide you a batch.

It appears here that two of those tensors were a mismatch. My best guess would be that the images in your dataset are not all the same sizes. Apparently one of them might be 25x20 and the other is 27x23. Using a resize and/or crop transformation should help here.

3

Could you tell me how should I apply that?
Thank you