Question

const outer_1 = new Promise(resolve => {
  const inner_1 = new Promise(resolve1 => {
    console.log(1);
    resolve1();
  });
  const inner_2 = inner_1.then(() => {
    console.log(2);
  });
  const inner_3 = inner_2.then(() => {
    console.log(3);
  })
  resolve()
});
const outer_2 = outer_1.then(() => {
  console.log(4)
})
/*
And the execution order:

# sync
  inner_1 ->
    log(1)
    queue microtask (inner2)
  outer_1 ->
    queue microtask (outer2)
# microtask-checkpoint
  inner_2 -> 
    log(2)
    queue microtask (inner3)
  outer_2 ->
    log(4)
  inner_3 ->
    log(3)
*/

这是我的自定义数据集类，我试图递归地将目录中的每个h5文件加载。

我认为def getitem 出了点问题，但是我不确定这是什么。

当我尝试加载时

class HDF5Dataset(torch.utils.data.Dataset):

    def __setup_files(self):
        files = glob.glob(os.path.join(self.dir_path,'**/*.h5'))
        return files

    def __init__(self, dir_path, IMG_SIZE):
        self.dir_path = dir_path
        self.IMG_SIZE = IMG_SIZE
        self.files = self.__setup_files()
        self.length = len(self.files)
        self.transform = transforms.Compose([
        transforms.Resize((IMG_SIZE, IMG_SIZE)),
        transforms.RandomHorizontalFlip(p=0.5),
        transforms.ToTensor(),
        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
    ])

    def __getitem__(self, idx):
        record = self.files[idx]
        h5 = h5py.File(record , 'r')
        image = h5['data'].value
        label = h5['label'].value
        h5.close()
        image = np.asarray(image)
        image = Image.fromarray(image.astype('uint8'), 'RGB')

        return self.transform(image), label

    def __len__(self):
        return self.length

和此代码

dataloaders['train'] = torch.utils.data.DataLoader(datasets['train'],
                                              batch_size=batch_size, shuffle=True, pin_memory=True, 
                                              num_workers=12)

给出类型错误：

inputs, classes = next(iter(dataloaders['train']))

我想确切地了解如何在pytorch中为h5文件构建自定义数据集以及如何加载它们。

谢谢！

.h5文件的自定义pytorch数据集类中出了点问题

0 个答案: