TypeError: default_collate: batch must contain tensors, numpy arrays, numbers, dicts or lists; found object, but the output is torch.Tensor

1

Hi everyone, I’m trying to do some classification for the CelebA dataset, but I’m encountering an error with trying to load my data.

The code is:

class MultiClassCelebA(Dataset):
    
    def __init__(self, dataframe, folder_dir, transform=None, target_transform=None):
        
        self.dataframe = dataframe
        self.folder_dir = folder_dir
        self.transform = transform
        self.target_transform = target_transform
        self.file_names = dataframe.index
        self.labels = dataframe.labels.values.tolist()
        
    def __len__(self):
        return len(self.dataframe)
    
    def __getitem__(self, index):
        
        image = Image.open(os.path.join(self.folder_dir, self.file_names[index]))
        label = self.labels[index][0]
        if self.transform is not None:
            image = self.transform(image)
        if self.target_transform is not None:
            label = self.target_transform(label)
        return image, label

tfms = transforms.Compose([transforms.Resize((256, 256)),
                           transforms.PILToTensor()])

train_dl = MultiClassCelebA(train_df, celeb_path + '/train/', transform=tfms)
val_dl = MultiClassCelebA(val_df, celeb_path + '/val/', transform=tfms)

train_dataloader = DataLoader(train_dl, shuffle = True, batch_size = 32)
val_dataloader = DataLoader(val_dl, shuffle = True, batch_size = 32)
next(iter(train_dataloader))

but I’m getting the above TypeError from the last line. I’ve checked the output of the dataset, e.g. for index 2

type(train_dl[2][0])

and it confirms that it is a torch.Tensor. Why is the dataloader seeing it as an Object and how can I fix this? Thanks

2

I’m unsure what might be causing the issue. Creating a random PIL.Image inside the Dataset works fine:

class MultiClassCelebA(Dataset):
    def __init__(self, dataframe, folder_dir, transform=None, target_transform=None):
        self.transform = transform
        self.target_transform = target_transform
        
    def __len__(self):
        return 10
    
    def __getitem__(self, index):
        image = transforms.ToPILImage()(torch.randn(3, 224, 224))
        label = torch.randint(0, 10, (1,))
        if self.transform is not None:
            image = self.transform(image)
        if self.target_transform is not None:
            label = self.target_transform(label)
        return image, label

tfms = transforms.Compose([transforms.Resize((256, 256)),
                           transforms.PILToTensor()])

train_dl = MultiClassCelebA(None, None, transform=tfms)

train_dataloader = DataLoader(train_dl, shuffle = True, batch_size = 32)
next(iter(train_dataloader))
3

Fixed the issue in getitem by return image, label.float()