def __call__(self, kspace, target, attrs, fname, slice):
kspace = np.array(kspace)
target = np.array(target)
kspace = transforms.to_tensor(kspace)
target = transforms.to_tensor(target)
kspace = transforms.complex_center_crop(
kspace, (self.resolution, self.resolution))
target = transforms.center_crop(target,
(self.resolution, self.resolution))
return kspace, target
def _get_cutout_train_transform(self):
transform = torchvision.transforms.Compose([
transforms.normalize(self.mean, self.std),
transforms.cutout(self.config['cutout_size'],
self.config['cutout_prob'],
self.config['cutout_inside']),
transforms.to_tensor(),
])
return transform
def _get_cutout_train_transform(self):
transform = torchvision.transforms.Compose([
torchvision.transforms.RandomCrop(28, padding=4),
torchvision.transforms.RandomHorizontalFlip(),
transforms.normalize(self.mean, self.std),
transforms.cutout(self.config['cutout_size'],
self.config['cutout_prob'],
self.config['cutout_inside']),
transforms.to_tensor(),
])
return transform
def _get_random_erasing_train_transform(self):
transform = torchvision.transforms.Compose([
transforms.normalize(self.mean, self.std),
transforms.random_erasing(
self.config['random_erasing_prob'],
self.config['random_erasing_area_ratio_range'],
self.config['random_erasing_min_aspect_ratio'],
self.config['random_erasing_max_attempt']),
transforms.to_tensor(),
])
return transform
def verify_img_data(img_data, expected_output, mode):
if mode is None:
img = transforms.ToPILImage()(img_data)
assert img.mode == 'RGB' # default should assume RGB
else:
img = transforms.ToPILImage(mode=mode)(img_data)
assert img.mode == mode
split = img.split()
for i in range(3):
assert np.allclose(expected_output[i].numpy(),
transforms.to_tensor(split[i]).numpy())
def __call__(self, pic):
"""
Args:
pic (PIL or numpy.ndarray): Image to be converted to tensor
Returns:
Tensor: Converted image.
"""
if isinstance(pic, np.ndarray):
# This is what TorchVision 0.2.0 returns for transforms.toTensor() for np.ndarray
return torch.from_numpy(pic.transpose((2, 0, 1))).float().div(255)
else:
return transforms.to_tensor(pic)
def _get_random_erasing_train_transform(self):
transform = torchvision.transforms.Compose([
torchvision.transforms.Resize((32,32)),
torchvision.transforms.ColorJitter(0.1,0.1,0.1),
# torchvision.transforms.RandomRotation(15),
# RandomAffine(degrees=15,scale=(0.8,1.2),shear=15),
torchvision.transforms.RandomCrop(32, padding=4),
torchvision.transforms.RandomHorizontalFlip(),
transforms.normalize(self.mean, self.std),
transforms.random_erasing(
self.config['random_erasing_prob'],
self.config['random_erasing_area_ratio_range'],
self.config['random_erasing_min_aspect_ratio'],
self.config['random_erasing_max_attempt']),
transforms.to_tensor(),
])
return transform
def _get_test_transform(self):
transform = torchvision.transforms.Compose([
transforms.normalize(self.mean, self.std),
transforms.to_tensor(),
])
return transform
def _add_to_tensor(self):
self._train_transforms.append(transforms.to_tensor())
