def test_load_celeba_dataset(self):
sizes = [64, 128]
for size in sizes:
dataset = data_utils.load_celeba_dataset(size=size,
root=self.dataset_dir,
download=False)
img = dataset[0][0]
assert img.shape == (3, size, size)
def get_celeba_images_with_attr(attr,
root='./dataset',
size=128,
num_samples=None,
**kwargs):
"""
Loads sampled CelebA images with index.
Args:
index (ndarray): The index of images.
root (str): The root directory where all datasets are stored.
size (int): Size of image to resize to.
Returns:
ndarray: Batch of len(index) images in np array form.
"""
dataset = data_utils.load_celeba_dataset(
root=root,
size=size,
transform_data=True,
convert_tensor=False, # Prevents normalization.
**kwargs)
dataset_dir = os.path.join(root, 'celeba')
attr_index, not_attr_index = get_celeba_index_with_attr(dataset_dir, attr)
print(f'Number of images with attribute {len(attr_index)}')
print(f'Number of images without attribute {len(not_attr_index)}')
if num_samples:
if len(attr_index) > num_samples:
attr_index = np.random.choice(attr_index,
size=num_samples,
replace=False)
if len(not_attr_index) > num_samples:
not_attr_index = np.random.choice(not_attr_index,
size=num_samples,
replace=False)
attr_images = get_index_images(dataset, attr_index)
not_attr_images = get_index_images(dataset, not_attr_index)
print(f'Number of images with attribute {attr}: {len(attr_images)}')
print(f'Number of images without attribute {attr}: {len(not_attr_images)}')
return attr_images, not_attr_images
def get_celeba_images(num_samples, root='./datasets', size=128, **kwargs):
"""
Loads randomly sampled CelebA images.
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_celeba_dataset(
root=root,
size=size,
transform_data=True,
convert_tensor=False, # Prevents normalization.
**kwargs)
images = get_random_images(dataset, num_samples)
return images
def get_celeba_images_with_index(index, root='./dataset', size=128, **kwargs):
"""
Loads sampled CelebA images with index.
Args:
index (ndarray): The index of images.
root (str): The root directory where all datasets are stored.
size (int): Size of image to resize to.
Returns:
ndarray: Batch of len(index) images in np array form.
"""
dataset = data_utils.load_celeba_dataset(
root=root,
size=size,
transform_data=True,
convert_tensor=False, # Prevents normalization.
**kwargs)
images = get_index_images(dataset, index)
return images
def get_celeba_with_attr(attr,
root='./dataset',
split='train',
size=128,
**kwargs):
"""
Loads CelebA with attribute target.
Args:
attr (str): The name of attribute.
root (str): The root directory where all datasets are stored.
split (str): The split of data to use.
size (int): Size of image to resize to.
Returns:
ndarray: Batch of len(index) images in np array form.
"""
celeba_attr = {
'5_o_Clock_Shadow': 0,
'Arched_Eyebrows': 1,
'Attractive': 2,
'Bags_Under_Eyes': 3,
'Bald': 4,
'Bangs': 5,
'Big_Lips': 6,
'Big_Nose': 7,
'Black_Hair': 8,
'Blond_Hair': 9,
'Blurry': 10,
'Brown_Hair': 11,
'Bushy_Eyebrows': 12,
'Chubby': 13,
'Double_Chin': 14,
'Eyeglasses': 15,
'Goatee': 16,
'Gray_Hair': 17,
'Heavy_Makeup': 18,
'High_Cheekbones': 19,
'Male': 20,
'Mouth_Slightly_Open': 21,
'Mustache': 22,
'Narrow_Eyes': 23,
'No_Beard': 24,
'Oval_Face': 25,
'Pale_Skin': 26,
'Pointy_Nose': 27,
'Receding_Hairline': 28,
'Rosy_Cheeks': 29,
'Sideburns': 30,
'Smiling': 31,
'Straight_Hair': 32,
'Wavy_Hair': 33,
'Wearing_Earrings': 34,
'Wearing_Hat': 35,
'Wearing_Lipstick': 36,
'Wearing_Necklace': 37,
'Wearing_Necktie': 38,
'Young': 39
}
dataset = data_utils.load_celeba_dataset(
root=root,
split=split,
size=size,
transform_data=True,
convert_tensor=True,
target_transform=transforms.Lambda(
lambda a: 1 if a[celeba_attr[attr]] == 1 else 0),
**kwargs)
return dataset