Sizes of tensors must match except in dimension 0. Got 3 and 1 in dimension 1

[Dayananda-V]

Encounter training phase problem with Places2 256 size dataset , looking forward help to solve this.

Traceback (most recent call last):
File "train.py", line 2, in
main(mode=1)
File "/home/user/workspace/user/performance/edge-informed-sisr/main.py", line 50, in main
model.train()
File "/home/user/workspace/user/performance/edge-informed-sisr/src/edge_match.py", line 150, in train
self.sample()
File "/home/user/workspace/user/performance/edge-informed-sisr/src/edge_match.py", line 252, in sample
items = next(self.sample_iterator)
File "/home/user/workspace/user/performance/edge-informed-sisr/src/dataset.py", line 146, in create_iterator
for item in sample_loader:
File "/home/user/anaconda3/envs/user_workspace/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 346, in next
data = self._dataset_fetcher.fetch(index) # may raise StopIteration
File "/home/user/anaconda3/envs/user_workspace/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 47, in fetch
return self.collate_fn(data)
File "/home/user/anaconda3/envs/user_workspace/lib/python3.6/site-packages/torch/utils/data/_utils/collate.py", line 79, in default_collate
return [default_collate(samples) for samples in transposed]
File "/home/user/anaconda3/envs/user_workspace/lib/python3.6/site-packages/torch/utils/data/_utils/collate.py", line 79, in
return [default_collate(samples) for samples in transposed]
File "/home/user/anaconda3/envs/user_workspace/lib/python3.6/site-packages/torch/utils/data/_utils/collate.py", line 55, in default_collate
return torch.stack(batch, 0, out=out)
RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 0. Got 3 and 1 in dimension 1 at /opt/conda/conda-bld/pytorch_1570710743984/work/aten/src/TH/generic/THTensor.cpp:689

Note : This problem encounter while evaluation dataset not training dataset

[richardburleigh]

It means at least one of your images isn't formatted correctly (eg. greyscale).

Here's a hacked together script I made to find and move those images to another folder. Just be aware that I haven't done any checks for filenames, so any other file in the directory will be moved as well.

import os
from scipy.misc import imread
from tqdm import tqdm

directory = 'INDIRECTORY'
movedir = 'OUTDIRECTORY'

for filename in tqdm(os.listdir(directory)):
        file = directory + filename

        try:
            im = imread(file)
            #print(filename + " is " + str(im.shape))
            if im.shape != (1024,1024,3):
                try:
                    os.rename(file, movedir + filename)
                except:
                    continue
                print(filename + " is " + str(im.shape))
        except:
            print("Failed to load " + str(filename))
            try:
                os.rename(file, movedir + filename)
            except:
                continue