如何对样本和标签做相同的transform变换

例如，在语义分割和边缘检测中，输入图像和目标真值都是2D图像，必须对输入图像和目标真值执行相同的变换，那么如何实现呢？下面是一种实现方式，供参考！

import torchvision.transforms.functional as TF

class MyDataset(Dataset):
    def __init__(self, image_paths, target_paths, train=True):
        self.image_paths = image_paths
        self.target_paths = target_paths

    def transform(self, image, mask):
        # Resize
        resize = transforms.Resize(size=(520, 520))
        image = resize(image)
        mask = resize(mask)

        # Random crop
        i, j, h, w = transforms.RandomCrop.get_params(
            image, output_size=(512, 512))
        image = TF.crop(image, i, j, h, w)
        mask = TF.crop(mask, i, j, h, w)

        # Random horizontal flipping
        if random.random() > 0.5:
            image = TF.hflip(image)
            mask = TF.hflip(mask)

        # Random vertical flipping
        if random.random() > 0.5:
            image = TF.vflip(image)
            mask = TF.vflip(mask)

        # Transform to tensor
        image = TF.to_tensor(image)
        mask = TF.to_tensor(mask)
        return image, mask

    def __getitem__(self, index):
        image = Image.open(self.image_paths[index])
        mask = Image.open(self.target_paths[index])
        x, y = self.transform(image, mask)
        return x, y

    def __len__(self):
        return len(self.image_paths)

参考文献 https://discuss.pytorch.org/t/torchvision-transfors-how-to-perform-identical-transform-on-both-image-and-target/10606/15