def test_load_fake_dataset(self):
H, W, C = 32, 32, 3
dataset = data_utils.load_fake_dataset(root=self.dataset_dir,
image_size=(C, H, W))
img = dataset[0][0]
assert img.shape == (C, H, W)
def test_load_fake_dataset(self):
H, W, C = 32, 32, 3
for transform_data in [True, False]:
dataset = data_utils.load_fake_dataset(
root=self.dataset_dir,
image_size=(C, H, W),
transform_data=transform_data)
img = dataset[0][0]
if transform_data:
assert img.shape == (C, H, W)
else:
img = np.asarray(img)
assert img.shape == (32, 32, 3
) # no resizing done, default 32x32.
def get_fake_data_images(num_samples, root='./datasets', size=32, **kwargs):
"""
Loads fake images, especially for testing.
Args:
num_samples (int): The number of images to randomly sample.
root (str): The root directory where all datasets are stored.
size (int): Size of image to resize to.
Returns:
Tensor: Batch of num_samples images in np array form.
"""
dataset = data_utils.load_fake_dataset(
root=root,
image_size=(3, size, size),
transform_data=True,
convert_tensor=False, # Prevents normalization.
**kwargs)
images = get_random_images(dataset, num_samples)
return images