def test_transform_pipeline_serialization(seed, image, mask):
aug = A.Compose([
A.OneOrOther(
A.Compose([
A.Resize(1024, 1024),
A.RandomSizedCrop(min_max_height=(256, 1024), height=512, width=512, p=1),
A.OneOf([
A.RandomSizedCrop(min_max_height=(256, 512), height=384, width=384, p=0.5),
A.RandomSizedCrop(min_max_height=(256, 512), height=512, width=512, p=0.5),
])
]),
A.Compose([
A.Resize(1024, 1024),
A.RandomSizedCrop(min_max_height=(256, 1025), height=256, width=256, p=1),
A.OneOf([
A.HueSaturationValue(p=0.5),
A.RGBShift(p=0.7)
], p=1),
])
),
A.HorizontalFlip(p=1),
A.RandomBrightnessContrast(p=0.5)
])
serialized_aug = A.to_dict(aug)
deserialized_aug = A.from_dict(serialized_aug)
random.seed(seed)
aug_data = aug(image=image, mask=mask)
random.seed(seed)
deserialized_aug_data = deserialized_aug(image=image, mask=mask)
assert np.array_equal(aug_data['image'], deserialized_aug_data['image'])
assert np.array_equal(aug_data['mask'], deserialized_aug_data['mask'])
def test_force_apply():
"""
Unit test for https://github.com/albu/albumentations/issues/189
"""
aug = A.Compose(
[
A.OneOrOther(
A.Compose(
[
A.RandomSizedCrop(min_max_height=(256, 1025), height=512, width=512, p=1),
A.OneOf(
[
A.RandomSizedCrop(min_max_height=(256, 512), height=384, width=384, p=0.5),
A.RandomSizedCrop(min_max_height=(256, 512), height=512, width=512, p=0.5),
]
),
]
),
A.Compose(
[
A.RandomSizedCrop(min_max_height=(256, 1025), height=256, width=256, p=1),
A.OneOf([A.HueSaturationValue(p=0.5), A.RGBShift(p=0.7)], p=1),
]
),
),
A.HorizontalFlip(p=1),
A.RandomBrightnessContrast(p=0.5),
]
)
res = aug(image=np.zeros((1248, 1248, 3), dtype=np.uint8))
assert res["image"].shape[0] in (256, 384, 512)
assert res["image"].shape[1] in (256, 384, 512)
def test_transform_pipeline_serialization_with_keypoints(seed, image, keypoints, keypoint_format, labels):
aug = A.Compose([
A.OneOrOther(
A.Compose([
A.RandomRotate90(),
A.OneOf([
A.HorizontalFlip(p=0.5),
A.VerticalFlip(p=0.5),
])
]),
A.Compose([
A.Rotate(p=0.5),
A.OneOf([
A.HueSaturationValue(p=0.5),
A.RGBShift(p=0.7)
], p=1),
])
),
A.HorizontalFlip(p=1),
A.RandomBrightnessContrast(p=0.5)
], keypoint_params={'format': keypoint_format, 'label_fields': ['labels']})
serialized_aug = A.to_dict(aug)
deserialized_aug = A.from_dict(serialized_aug)
random.seed(seed)
aug_data = aug(image=image, keypoints=keypoints, labels=labels)
random.seed(seed)
deserialized_aug_data = deserialized_aug(image=image, keypoints=keypoints, labels=labels)
assert np.array_equal(aug_data['image'], deserialized_aug_data['image'])
assert np.array_equal(aug_data['keypoints'], deserialized_aug_data['keypoints'])
def test_transform_pipeline_serialization_with_bboxes(seed, image, bboxes,
bbox_format, labels):
aug = A.Compose(
[
A.OneOrOther(
A.Compose([
A.RandomRotate90(),
A.OneOf([A.HorizontalFlip(p=0.5),
A.VerticalFlip(p=0.5)])
]),
A.Compose([
A.Rotate(p=0.5),
A.OneOf([A.HueSaturationValue(p=0.5),
A.RGBShift(p=0.7)],
p=1)
]),
),
A.HorizontalFlip(p=1),
A.RandomBrightnessContrast(p=0.5),
],
bbox_params={
"format": bbox_format,
"label_fields": ["labels"]
},
)
serialized_aug = A.to_dict(aug)
deserialized_aug = A.from_dict(serialized_aug)
set_seed(seed)
aug_data = aug(image=image, bboxes=bboxes, labels=labels)
set_seed(seed)
deserialized_aug_data = deserialized_aug(image=image,
bboxes=bboxes,
labels=labels)
assert np.array_equal(aug_data["image"], deserialized_aug_data["image"])
assert np.array_equal(aug_data["bboxes"], deserialized_aug_data["bboxes"])
def create_runtime_tfms():
mean, std = [0.5] * 3, [0.5] * 3
resize_to_64 = A.SmallestMaxSize(IMG_SIZE, interpolation=cv2.INTER_AREA)
out = [
A.HorizontalFlip(p=0.5),
A.Normalize(mean=mean, std=std),
ToTensor()
]
rand_crop = A.Compose([
A.SmallestMaxSize(IMG_SIZE + 8, interpolation=cv2.INTER_AREA),
A.RandomCrop(IMG_SIZE, IMG_SIZE)
])
affine_1 = A.ShiftScaleRotate(shift_limit=0,
scale_limit=0.1,
rotate_limit=8,
interpolation=cv2.INTER_CUBIC,
border_mode=cv2.BORDER_REFLECT_101,
p=1.0)
affine_1 = A.Compose([affine_1, resize_to_64])
affine_2 = A.ShiftScaleRotate(shift_limit=0.06,
scale_limit=(-0.06, 0.18),
rotate_limit=6,
interpolation=cv2.INTER_CUBIC,
border_mode=cv2.BORDER_REFLECT_101,
p=1.0)
affine_2 = A.Compose([affine_2, resize_to_64])
tfm_0 = A.Compose(out)
tfm_1 = A.Compose([A.OneOrOther(affine_1, rand_crop, p=1.0), *out])
tfm_2 = A.Compose([affine_2, *out])
return [tfm_0, tfm_1, tfm_2]
def aug_tfms(
size: Union[int, Tuple[int, int]],
presize: Optional[Union[int, Tuple[int, int]]] = None,
horizontal_flip: Optional[A.HorizontalFlip] = A.HorizontalFlip(),
shift_scale_rotate: Optional[A.ShiftScaleRotate] = A.ShiftScaleRotate(),
rgb_shift: Optional[A.RGBShift] = A.RGBShift(),
lightning: Optional[A.RandomBrightnessContrast] = A.RandomBrightnessContrast(),
blur: Optional[A.Blur] = A.Blur(blur_limit=(1, 3)),
crop_fn: Optional[A.DualTransform] = partial(A.RandomSizedBBoxSafeCrop, p=0.5),
pad: Optional[A.DualTransform] = partial(
A.PadIfNeeded, border_mode=cv2.BORDER_CONSTANT, value=[124, 116, 104]
),
) -> List[A.BasicTransform]:
"""Collection of useful augmentation transforms.
# Arguments
size: The final size of the image. If an `int` is given, the maximum size of
the image is rescaled, maintaing aspect ratio. If a `tuple` is given,
the image is rescaled to have that exact size (height, width).
presizing: Rescale the image before applying other transfroms. If `None` this
transform is not applied. First introduced by fastai,this technique is
explained in their book in [this](https://github.com/fastai/fastbook/blob/master/05_pet_breeds.ipynb)
chapter (tip: search for "Presizing").
horizontal_flip: Flip around the y-axis. If `None` this transform is not applied.
shift_scale_rotate: Randomly shift, scale, and rotate. If `None` this transform
is not applied.
rgb_shift: Randomly shift values for each channel of RGB image. If `None` this
transform is not applied.
lightning: Randomly changes Brightness and Contrast. If `None` this transform
is not applied.
blur: Randomly blur the image. If `None` this transform is not applied.
crop_fn: Randomly crop the image. If `None` this transform is not applied.
Use `partial` to saturate other parameters of the class.
pad: Pad the image to `size`, squaring the image if `size` is an `int`.
If `None` this transform is not applied. Use `partial` to sature other
parameters of the class.
# Returns
A list of albumentations transforms.
"""
height, width = (size, size) if isinstance(size, int) else size
tfms = []
tfms += [_resize(presize, A.SmallestMaxSize) if presize is not None else None]
tfms += [horizontal_flip, shift_scale_rotate, rgb_shift, lightning, blur]
# Resize as the last transforms to reduce the number of artificial artifacts created
if crop_fn is not None:
crop = crop_fn(height=height, width=width)
tfms += [A.OneOrOther(crop, _resize(size), p=crop.p)]
else:
tfms += [_resize(size)]
tfms += [pad(min_height=height, min_width=width) if pad is not None else None]
tfms = [tfm for tfm in tfms if tfm is not None]
return tfms
def crop_transform(image_size: Tuple[int, int], min_scale=0.75, max_scale=1.25, input_size=5000):
return A.OneOrOther(
A.RandomSizedCrop(
(int(image_size[0] * min_scale), int(min(input_size, image_size[0] * max_scale))),
image_size[0],
image_size[1],
),
A.CropNonEmptyMaskIfExists(image_size[0], image_size[1]),
)
def crop_transform_xview2(image_size: Tuple[int, int], min_scale=0.4, max_scale=0.75, input_size=1024):
return A.OneOrOther(
A.RandomSizedCrop(
(int(image_size[0] * min_scale), int(min(input_size, image_size[0] * max_scale))),
image_size[0],
image_size[1],
),
A.Compose(
[A.Resize(input_size * 2, input_size * 2), A.CropNonEmptyMaskIfExists(image_size[0], image_size[1])]
),
)
def set_transforms(self) -> dict:
d4_tansforms = [
A.SmallestMaxSize(self.img_size[0], interpolation=0, p=1.),
# D4 Group augmentations
A.HorizontalFlip(p=0.5),
A.VerticalFlip(p=0.5),
A.RandomRotate90(p=0.5),
A.Transpose(p=0.5),
]
tensor_norm = [
ToTensor(),
Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
]
geometric = [
A.SmallestMaxSize(self.img_size[0], interpolation=0, p=1.),
A.ShiftScaleRotate(shift_limit=self.shift_limit,
self.scale_limit,
rotate_limit=self.rotate_limit,
interpolation=cv2.INTER_LINEAR,
border_mode=cv2.BORDER_CONSTANT,
value=0,
mask_value=0,
p=0.5),
# crop and resize
A.RandomSizedCrop(
(self.crop_size[0],
min(self.crop_size[1], self.img_size[0], self.img_size[1])),
self.img_size[0],
self.img_size[1],
w2h_ratio=1.0,
interpolation=cv2.INTER_LINEAR,
always_apply=False,
p=0.2),
]
resize_crop = [
A.SmallestMaxSize(max(self.img_size[0], self.img_size[1]),
interpolation=0,
p=1.),
A.RandomCrop(self.img_size[0], self.img_size[1], p=1.),
]
train_light = [
A.SmallestMaxSize(self.img_size[0], interpolation=0, p=1.),
A.RandomCrop(self.img_size[0], self.img_size[1], p=1.),
A.HorizontalFlip(p=0.5),
A.RandomBrightnessContrast(p=0.5),
]
train_medium = train_light.append([
A.ShiftScaleRotate(shift_limit=self.shift_limit,
self.scale_limit,
rotate_limit=self.rotate_limit,
p=0.5),
A.OneOf([
A.RandomBrightnessContrast(p=0.7),
A.Equalize(p=0.3),
A.HueSaturationValue(p=0.5),
A.RGBShift(p=0.5),
A.RandomGamma(p=0.4),
A.ChannelShuffle(p=0.05),
],
p=0.9),
A.OneOf([
A.GaussNoise(p=0.5),
A.ISONoise(p=0.5),
A.MultiplicativeNoise(0.5),
],
p=0.2),
])
valid_ade = [
A.SmallestMaxSize(self.img_size, p=1.),
A.Lambda(name="Pad32", image=pad_x32, mask=pad_x32),
]
# from bloodaxe
# https://github.com/BloodAxe/Catalyst-Inria-Segmentation-Example/blob/master/inria/augmentations.py
crop = [ #(image_size: Tuple[int, int], min_scale=0.75, max_scale=1.25, input_size=5000):
A.OneOrOther(
A.RandomSizedCrop((self.crop_size[0],
min(self.crop_size[1], self.img_size[0],
self.img_size[1])), self.img_size[0],
self.img_size[1]),
A.CropNonEmptyMaskIfExists(self.img_size[0], self.img_size[1]),
)
]
safe_augmentations = [A.HorizontalFlip(), A.RandomBrightnessContrast()]
light_augmentations = [
A.HorizontalFlip(),
A.RandomBrightnessContrast(),
A.OneOf([
A.ShiftScaleRotate(scale_limit=self.scale_limit,
rotate_limit=self.rotate_limit,
border_mode=cv2.BORDER_CONSTANT),
A.IAAAffine(),
A.IAAPerspective(),
A.NoOp()
]),
A.HueSaturationValue(),
]
medium_augmentations = [
A.HorizontalFlip(),
A.ShiftScaleRotate(scale_limit=self.scale_limit,
rotate_limit=self.rotate_limit,
border_mode=cv2.BORDER_CONSTANT),
# Add occasion blur/sharpening
A.OneOf([A.GaussianBlur(),
A.IAASharpen(),
A.NoOp()]),
# Spatial-preserving augmentations:
A.OneOf(
[A.CoarseDropout(),
A.MaskDropout(max_objects=5),
A.NoOp()]),
A.GaussNoise(),
A.OneOf([
A.RandomBrightnessContrast(),
A.CLAHE(),
A.HueSaturationValue(),
A.RGBShift(),
A.RandomGamma()
]),
# Weather effects
A.RandomFog(fog_coef_lower=0.01, fog_coef_upper=0.3, p=0.1),
]
hard_augmentations = [
A.RandomRotate90(),
A.Transpose(),
A.RandomGridShuffle(),
A.ShiftScaleRotate(scale_limit=self.scale_limit,
rotate_limit=self.rotate_limit,
border_mode=cv2.BORDER_CONSTANT,
mask_value=0,
value=0),
A.ElasticTransform(border_mode=cv2.BORDER_CONSTANT,
alpha_affine=5,
mask_value=0,
value=0),
# Add occasion blur
A.OneOf([
A.GaussianBlur(),
A.GaussNoise(),
A.IAAAdditiveGaussianNoise(),
A.NoOp()
]),
# D4 Augmentations
A.OneOf(
[A.CoarseDropout(),
A.MaskDropout(max_objects=10),
A.NoOp()]),
# Spatial-preserving augmentations:
A.OneOf([
A.RandomBrightnessContrast(brightness_by_max=True),
A.HueSaturationValue(),
A.RGBShift(),
A.RandomGamma(),
A.NoOp(),
]),
# Weather effects
A.OneOf([
A.RandomFog(fog_coef_lower=0.01, fog_coef_upper=0.3, p=0.1),
A.NoOp()
]),
]
TRANSFORMS = {
"d4": d4_tansforms,
"resize_crop": resize_crop,
"geometric": geometric,
"light": train_light,
"medium": train_medium,
"ade_valid": valid_ade,
"flip_bright": safe_augmentations,
"crop": crop,
"inria_light": light_augmentations,
"inria_medium": medium_augmentations,
"inria_hard": hard_augmentations,
"inria_valid": safe_augmentations,
}
return TRANSFORMS
def get_transforms(self):
augs = []
if self.augs_name:
augs = self.set_transforms[self.augs_name]
augs.append(
A.Resize(height=self.img_size[0],
width=self.img_size[1],
p=1.0))
if self.use_d4:
augs.append(self.set_transforms["d4"])
if self.vflip:
augs.append(A.VerticalFlip(p=0.5))
if self.hflip:
augs.append(A.HorizontalFlip(p=0.5))
if self.normalise:
augs.append(A.Normalize())
return A.Compose(augs)
