Pytorch でシーケンスデータを順番で読込 ref: https://qiita.com/shuuchen/items/466dc7977a146f7f38f2

file0.txt

class ImageFolderWithPaths(datasets.ImageFolder):
    """Custom dataset that includes image file paths. Extends
    torchvision.datasets.ImageFolder
    """

    # override the __getitem__ method. this is the method dataloader calls
    def __getitem__(self, index):
        # this is what ImageFolder normally returns 
        original_tuple = super(ImageFolderWithPaths, self).__getitem__(index)
        # the image file path
        path = self.imgs[index][0]
        # make a new tuple that includes original and the path
        tuple_with_path = (original_tuple + (path,))
        return tuple_with_path

file1.txt

data_dir = './pregnant'

data_transforms = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),])

image_datasets = {x: ImageFolderWithPaths(os.path.join(data_dir, x), 
                                          transform=data_transforms) for x in ['all']}

data_loaders = {x: torch.utils.data.DataLoader(image_datasets[x], 
                                               batch_size=batch_size, shuffle=False) for x in ['all']}

dataset_sizes = {x: len(image_datasets[x]) for x in ['all']}

for inputs, _, paths in data_loaders['all']:
    print(paths)
    break

file2.txt

'0.jpg', 
'1.jpg', 
'10.jpg', 
'100.jpg', 
'1000.jpg', 
'1001.jpg',
'1002.jpg', 
'1003.jpg', 
'1004.jpg', 
'1005.jpg', 
'1006.jpg', 
'1007.jpg'
...

file3.txt

from PIL import Image
data_iter = iter(data_loaders['all'])

# 本格
for i in range(1488 - batch_size):  
    imgs = []
    for ii in range(i, i + batch_size):
        path = os.path.join('{}.jpg'.format(ii)) 
        print(path)
        img = data_transforms(Image.open(path))
        imgs.append(img)
    print(len(imgs))
    imgs = torch.stack(imgs)
    print(imgs.size())
    break

# 比較用
for inputs, _, paths in data_loaders['all']:
    print(inputs.size())
    break

file4.txt

0.jpg
1.jpg
2.jpg
3.jpg
4.jpg
5.jpg
6.jpg
7.jpg
8.jpg
...
36
torch.Size([36, 3, 224, 224])
torch.Size([36, 3, 224, 224])