def chapter_augmenters_cutout():
aug = iaa.Cutout(nb_iterations=2)
run_and_save_augseq("arithmetic/cutout_nb_iterations_2.jpg",
aug,
[ia.quokka(size=(128, 128)) for _ in range(4 * 2)],
cols=4,
rows=2)
aug = iaa.Cutout(nb_iterations=(1, 5), size=0.2, squared=False)
run_and_save_augseq("arithmetic/cutout_non_square.jpg",
aug,
[ia.quokka(size=(128, 170)) for _ in range(3 * 2)],
cols=3,
rows=2)
aug = iaa.Cutout(fill_mode="constant", cval=255)
run_and_save_augseq("arithmetic/cutout_cval_255.jpg",
aug,
[ia.quokka(size=(128, 128)) for _ in range(4 * 2)],
cols=4,
rows=2)
aug = iaa.Cutout(fill_mode="constant", cval=(0, 255), fill_per_channel=0.5)
run_and_save_augseq("arithmetic/cutout_rgb.jpg",
aug,
[ia.quokka(size=(128, 128)) for _ in range(4 * 2)],
cols=4,
rows=2)
aug = iaa.Cutout(fill_mode="gaussian", fill_per_channel=True)
run_and_save_augseq("arithmetic/cutout_gaussian.jpg",
aug,
[ia.quokka(size=(128, 128)) for _ in range(4 * 2)],
cols=4,
rows=2)
def __init__(self,num_of_augms=0):
self.num_of_augms=num_of_augms
self.aug=iaa.OneOf([
iaa.Sequential([
iaa.LinearContrast(alpha=(0.75, 1.5)),
iaa.Fliplr(0.5)
]),
iaa.Sequential([
iaa.Grayscale(alpha=(0.1, 0.9)),
iaa.Affine(
translate_percent={"y": (-0.15, 0.15)}
)
]),
iaa.Sequential([
iaa.LinearContrast((0.6, 1.4)),
iaa.ShearX((-10, 10))
]),
iaa.Sequential([
iaa.GaussianBlur(sigma=(0, 1)),
iaa.ShearY((-10, 10))
]),
iaa.Sequential([
iaa.Cutout(nb_iterations=(1, 2), size=0.1, squared=False),
iaa.Multiply((0.8, 1.2), per_channel=0.25),
iaa.Fliplr(0.5),
]),
iaa.Sequential([
iaa.LinearContrast((0.6, 1.4)),
iaa.Affine(
translate_percent={"x": (-0.25, 0.25)}
)
]),
iaa.Sequential([
iaa.Cutout(nb_iterations=(1, 5), size=0.1, squared=False),
iaa.AdditiveGaussianNoise(loc=0, scale=(0.0, 15), per_channel=0.5),
iaa.Affine(
scale={"x": (0.9, 1.1), "y": (0.9, 1.1)},
)
]),
iaa.Sequential([
iaa.CoarseDropout((0.0, 0.05), size_percent=(0.02, 0.25)),
iaa.GaussianBlur(sigma=(0, 2)),
iaa.Affine(
scale={"x": (0.8, 1.2), "y": (0.8, 1.2)}
)])
# iaa.Sequential([
# iaa.Cutout(nb_iterations=(1, 5), size=0.05, squared=False),
# iaa.Grayscale(alpha=(0.0, 0.50)),
# iaa.ScaleX((0.75, 1.25))
# ]),
# iaa.Sequential([
# iaa.LinearContrast((0.8, 1.2), per_channel=True),
# iaa.PerspectiveTransform(scale=(0.01, 0.15))
# ])
])
def __call__(self, x, y, choice):
if choice:
if self.fillcolor:
aug = iaa.Cutout(size=0.5,
fill_mode='constant',
cval=(0, 255),
fill_per_channel=0.5)
else:
aug = iaa.Cutout(size=0.5, fill_mode='constant', cval=0)
x = aug.augment_image(x)
return x, y
def setup_augmentation(self):
# Augmentation
# aug = iaa.Sequential([
# #iaa.Sometimes(0.5, iaa.PerspectiveTransform(0.05)),
# #iaa.Sometimes(0.5, iaa.CropAndPad(percent=(-0.05, 0.1))),
# #iaa.Sometimes(0.5, iaa.Affine(scale=(1.0, 1.2))),
# iaa.Sometimes(0.5, iaa.CoarseDropout( p=0.05, size_percent=0.01) ),F
# iaa.Sometimes(0.5, iaa.GaussianBlur(1.2*np.random.rand())),
# iaa.Sometimes(0.5, iaa.Add((-0.1, 0.1), per_channel=0.3)),
# iaa.Sometimes(0.3, iaa.Invert(0.2, per_channel=True)),
# iaa.Sometimes(0.5, iaa.Multiply((0.6, 1.4), per_channel=0.5)),
# iaa.Sometimes(0.5, iaa.Multiply((0.6, 1.4))),
# iaa.Sometimes(0.5, iaa.ContrastNormalization((0.5, 2.2), per_channel=0.3))],
# random_order=False)
# aug = iaa.Sequential([
# #iaa.Sometimes(0.5, iaa.CoarseDropout( p=0.25, size_percent=0.02) ),
# iaa.Sometimes(0.5, iaa.GaussianBlur(1.2*np.random.rand())),
# iaa.Sometimes(0.5, iaa.Add((-60, 60), per_channel=0.3)),
# iaa.Sometimes(0.5, iaa.Multiply((0.6, 1.4), per_channel=0.5)),
# iaa.Sometimes(0.5, iaa.Multiply((0.6, 1.4))),
# iaa.Sometimes(0.5, iaa.ContrastNormalization((0.5, 2.2), per_channel=0.3))],
# random_order=False)
aug = iaa.Sequential(
[
#iaa.Sometimes(0.5, PerspectiveTransform(0.05)),
#iaa.Sometimes(0.5, CropAndPad(percent=(-0.05, 0.1))),
iaa.Sometimes(0.5, iaa.Affine(scale=(1.0, 1.2))),
#iaa.Sometimes(0.5, iaa.CoarseDropout( p=0.2, size_percent=0.05) ),
iaa.Sometimes(
0.5,
iaa.SomeOf(2, [
iaa.CoarseDropout(p=0.2, size_percent=0.05),
iaa.Cutout(fill_mode="constant",
cval=(0, 255),
fill_per_channel=0.5),
iaa.Cutout(fill_mode="constant", cval=(255)),
iaa.CoarseSaltAndPepper(0.05, size_px=(4, 16)),
iaa.CoarseSalt(0.05, size_percent=(0.01, 0.1))
])),
iaa.Sometimes(0.5, iaa.GaussianBlur(1.2 * np.random.rand())),
iaa.Sometimes(0.5, iaa.Add((-25, 25), per_channel=0.3)),
iaa.Sometimes(0.3, iaa.Invert(0.2, per_channel=True)),
iaa.Sometimes(0.5, iaa.Multiply((0.6, 1.4), per_channel=0.5)),
iaa.Sometimes(0.5, iaa.Multiply((0.6, 1.4))),
iaa.Sometimes(
0.5, iaa.ContrastNormalization(
(0.5, 2.2), per_channel=0.3))
],
random_order=False)
return aug
def main():
aug = iaa.Cutout(fill_mode=["gaussian", "constant"],
cval=(0, 255),
fill_per_channel=0.5)
image = ia.quokka()
images_aug = aug(images=[image] * 16)
ia.imshow(ia.draw_grid(images_aug, cols=4, rows=4))
def __call__(self, sample):
img, annot = sample.image, sample.annotation
unique_labels = np.unique(
annot[:, 4].astype('int').astype('str')).tolist()
bbs = BoundingBoxesOnImage([
BoundingBox(x1=ann[0],
y1=ann[1],
x2=ann[2],
y2=ann[3],
label=str(int(ann[4]))) for ann in annot
],
shape=img.shape)
aug = iaa.BlendAlphaBoundingBoxes(labels=unique_labels,
foreground=iaa.Cutout(
nb_iterations=int(round(self.v)),
size=0.05,
fill_mode="gaussian"))
img_aug, bbs_aug = aug(image=img, bounding_boxes=bbs)
annot_aug = np.array(
[[bb.x1, bb.y1, bb.x2, bb.y2,
np.float32(bb.label)] for bb in bbs_aug])
# the shape has to be at least (0,5)
if len(annot_aug) == 0:
annot_aug = np.zeros((0, 5))
sample.image = img_aug
sample.annotation = annot_aug
def __init__(self):
self.aug = iaa.Sequential([
iaa.Sometimes(
0.15,
iaa.OneOf([
iaa.GammaContrast(gamma=(0, 1.75)),
iaa.pillike.Autocontrast(cutoff=(0, 15.0))
])),
iaa.Sometimes(
0.15,
iaa.OneOf([
iaa.HistogramEqualization(),
iaa.pillike.Equalize(),
])),
iaa.Sometimes(0.1, iaa.Grayscale(alpha=(0.05, 1.0))),
iaa.Sometimes(0.2, iaa.JpegCompression(compression=(70, 99))),
iaa.Sometimes(0.1,
iaa.UniformColorQuantizationToNBits(nb_bits=(2, 8))),
iaa.Sometimes(
0.3,
iaa.MultiplyAndAddToBrightness(mul=(0.5, 1.5), add=(-30, 30))),
iaa.Sometimes(
0.2,
iaa.Cutout(
fill_mode="constant", cval=(0, 255), fill_per_channel=0.5))
],
random_order=True)
def cutout_bbox(magnitude: int, **kwargs) -> iaa.BlendAlphaBoundingBoxes:
"""
Only apply cutout to the bounding box area. Passing the
height and width of the image as integers and as keywords
will scale the bounding box according to the policy. Note, the
cutout location is chosen randomly and will only appear if it
falls within the bounding box.
:type magnitude: int
:param magnitude: magnitude of cutout
:param kwargs:
height: height of the image as int
width: width of the image as int
:rtype: iaa.BlendAlphaBoundingBoxes
:return: Method to apply cutout only to bounding boxes
"""
level = int((magnitude / _MAX_MAGNITUDE) * CUTOUT_BBOX)
cutout_args = {}
if 'height' in kwargs and 'width' in kwargs:
size = tuple(
np.clip([level / kwargs['height'], level / kwargs['width']], 0.0,
1.0))
cutout_args['size'] = size
return iaa.BlendAlphaBoundingBoxes(None,
foreground=iaa.Cutout(**cutout_args))
def __call__(self, img, labels, mode=None):
aug_labels = labels.copy()
if random.random() < self.p:
aug = iaa.Cutout(**self.params)
img = aug.augment_image(img)
return img, labels
def img_aug(img_name, base_dir):
image = mpimg.imread(Path(base_dir, 'datasets', 'images', img_name))
ran = random.randint(1, 5)
if ran == 1:
aug = iaa.Sequential([
iaa.CropAndPad(percent=(-0.2, 0.2), pad_mode="edge"),
iaa.AddToHueAndSaturation((-60, 60)),
iaa.ElasticTransformation(alpha=90, sigma=9),
iaa.Cutout()
],
random_order=True)
image = aug(image=image)
elif ran == 2:
aug = iaa.BlendAlphaRegularGrid(nb_rows=2,
nb_cols=2,
foreground=iaa.Multiply(0.0),
background=iaa.AveragePooling(8),
alpha=[0.0, 0.0, 1.0])
image = aug(image=image)
elif ran == 3:
image = tf.image.adjust_brightness(image, 0.4)
elif ran == 4:
image = tf.image.random_flip_left_right(image)
image = tf.image.rot90(image)
image = tf.image.random_flip_up_down(image)
elif ran == 5:
image = tf.image.central_crop(image, central_fraction=0.5)
return image
def cutout(magnitude: int, **kwargs) -> iaa.Cutout:
"""
Apply cutout anywhere in the image. Passing the height and width
of the image as integers and as keywords will scale the bounding
box according to the policy
Tensorflow Policy Equivalent: cutout
The cutout value in the policies is at a pixel level. The imgaug cutout
augmentation method requires the cutout to be a percentage of the image.
Passing the image height and width as kwargs will scale the cutout to
the appropriate percentage. Otherwise the imgaug default of 20% will be
used.
:type magnitude: int
:param magnitude: magnitude of cutout
:param kwargs:
height: height of the image as int
width: width of the image as int
:rtype: iaa.Cutout
:return: Method to apply cutout to image
"""
level = int((magnitude / _MAX_MAGNITUDE) * CUTOUT_CONST)
cutout_args = {}
if 'height' in kwargs and 'width' in kwargs:
size = tuple(
np.clip([(level / kwargs['height']) * 2,
(level / kwargs['width']) * 2], 0.0, 1.0))
cutout_args['size'] = size
return iaa.Cutout(**cutout_args)
def aug_image(image):
seq = iaa.Sequential([
iaa.Fliplr(0.5),
iaa.Affine(rotate= (-8, 8),
shear = (-8, 8),
mode='edge'),
iaa.SomeOf((0, 2),
[
iaa.GaussianBlur((0, 0.3)),
iaa.AdditiveGaussianNoise(loc=0, scale=(0.0, 0.01 * 255), per_channel=0.5),
iaa.AddToHueAndSaturation((-5, 5)), # change hue and saturation
iaa.PiecewiseAffine(scale=(0.01, 0.03)),
iaa.PerspectiveTransform(scale=(0.01, 0.1)),
iaa.JpegCompression(20, 40)
],
random_order=True
),
iaa.Cutout(nb_iterations=1, size=(0.02, 0.2), squared=False)
])
image = seq.augment_image(image)
return image
def __init__(self, agumentation={}):
try:
self.aug = iaa.Sequential([
iaa.Resize((224, 224)),
iaa.Sometimes(
float(agumentation['GaussianBlur_P']),
iaa.GaussianBlur(
sigma=(agumentation["GaussianBlur_sigma"]))),
iaa.Sometimes(
float(agumentation['MedianBlur_P']),
iaa.imgcorruptlike.MotionBlur(
severity=agumentation['MedianBlur_Severity'])),
iaa.Sometimes(
agumentation['DefocusBlur_P'],
iaa.imgcorruptlike.DefocusBlur(
severity=agumentation['DefocusBlur_Severity'])),
iaa.Sometimes(
agumentation['Cutout_P'],
iaa.Cutout(
nb_iterations=agumentation['Cutout_cum_of_cuts']),
),
iaa.ChannelShuffle(agumentation['ChannelSuffle_P']),
iaa.Sometimes(
agumentation['ElasticTransformation_P'],
iaa.ElasticTransformation(
alpha=agumentation['ElasticTransformation_alpha'],
sigma=agumentation['ElasticTransformation_sigma']),
),
iaa.Sometimes(
agumentation['PerspectiveTransform_P'],
iaa.PerspectiveTransform(
scale=agumentation['PerspectiveTransform_scale']),
),
],
random_order=True)
except:
self.aug = iaa.Sequential([
iaa.Resize((224, 224)),
iaa.Sometimes(0.25, iaa.GaussianBlur(sigma=(0, 3.0))),
iaa.Fliplr(0.5),
iaa.Affine(rotate=(-20, 20), mode='symmetric'),
iaa.Dropout2d(p=0.5),
iaa.Sometimes(0.5, iaa.Cutout(nb_iterations=2)),
iaa.Sometimes(0.25, iaa.Invert(0.25, per_channel=0.5)),
iaa.Sometimes(0.25, iaa.Add((-40, 40))),
iaa.AddToHueAndSaturation(value=(-10, 10), per_channel=True)
])
def __init__(self, nb_iterations=(3, 8), size=0.1, determint=False):
self.nb_iterations = nb_iterations
self.size = size
self.determint = determint
self.func = iaa.Cutout(nb_iterations=nb_iterations,
size=size,
squared=False,
fill_mode="gaussian")
def generate_cutout():
ia.seed(1)
image = ia.quokka((128, 128))
images_aug = []
images_aug.append(image)
images_aug.extend(iaa.Cutout()(images=[image] * 7))
images_aug.append(image)
images_aug.extend(
iaa.Cutout(nb_iterations=2,
cval=(0, 255),
fill_mode=["constant", "gaussian"],
fill_per_channel=0.8)(images=[image] * 7))
_save("cutout.jpg", ia.draw_grid(images_aug, cols=7, rows=2))
def get_preview(images, augmentationList):
"""
Accepts a list of images and augmentationList as input.
Provides a list of augmented images in that order as ouptut.
"""
augmented = []
for image in images:
for augmentation in augmentationList:
aug_id = augmentation['id']
params = augmentation['params']
if (aug_id == 1):
image = iaa.SaltAndPepper(p=params[0],
per_channel=params[1])(image=image)
elif (aug_id == 2):
image = iaa.imgcorruptlike.GaussianNoise(
severity=(params[0], params[1]))(image=image)
elif (aug_id == 3):
image = iaa.Rain(speed=(params[0], params[1]),
drop_size=(params[2], params[3]))(image=image)
elif (aug_id == 4):
image = iaa.imgcorruptlike.Fog(
severity=(params[0], params[1]))(image=image)
elif (aug_id == 5):
image = iaa.imgcorruptlike.Snow(
severity=(params[0], params[1]))(image=image)
elif (aug_id == 6):
image = iaa.imgcorruptlike.Spatter(
severity=(params[0], params[1]))(image=image)
elif (aug_id == 7):
image = iaa.BlendAlphaSimplexNoise(
iaa.EdgeDetect(1))(image=image)
elif (aug_id == 8):
image = iaa.Rotate(rotate=(params[0], params[1]))(image=image)
elif (aug_id == 9):
image = iaa.Affine()(image=image) #to be implemented
elif (aug_id == 10):
image = iaa.MotionBlur(k=params[0],
angle=(params[1],
params[2]))(image=image)
elif (aug_id == 11):
image = iaa.imgcorruptlike.ZoomBlur(
severity=(params[0], params[1]))(image=image)
elif (aug_id == 12):
image = iaa.AddToBrightness()(image=image) #to be implemented
elif (aug_id == 13):
image = iaa.ChangeColorTemperature(
kelvin=(params[0], params[1]))(image=image)
elif (aug_id == 14):
image = iaa.SigmoidContrast()(image=image) #to be implemented
elif (aug_id == 15):
image = iaa.Cutout(nb_iterations=(params[0], params[1]),
size=params[2],
squared=params[3])(image=image)
else:
print("Not implemented")
augmented.append(image)
return augmented
def fit(self, X: Dict[str, Any], y: Any = None) -> BaseImageAugmenter:
if self.use_augmenter:
self.augmenter: Augmenter = iaa.Sometimes(
self.p,
iaa.Cutout(nb_iterations=(1, 10),
size=(0.1, 0.5),
random_state=self.random_state),
name=self.get_properties()['name'])
return self
def __init__(self, train_file, class_list, batch_size=1, shuffle=True, transform=None, augment = False, crops_per_snapshot = 8, crop_size = (512,512), debug=False):
self.train_file = train_file
self.class_list = class_list
self.transform = transform
#region parse the provided class file
try:
with open(self.class_list, 'r', newline='') as file:
self.classes = self.load_classes(csv.reader(file, delimiter=','))
except ValueError as e:
raise (ValueError('invalid CSV class file: {}: {}'.format(self.class_list, e)))
self.labels = {}
for key, value in self.classes.items():
self.labels[value] = key
#endregion
#region parse annotations csv file
# lines should be of format "img_path, x1, y1, x2, y2, x3, y3, class_name"
try:
with open(self.train_file, 'r', newline='') as file:
self.image_data = self._read_annotations(csv.reader(file, delimiter=','), self.classes)
except ValueError as e:
raise (ValueError('invalid CSV annotations file: {}: {}'.format(self.train_file, e)))
#endregion
self.image_names = list(self.image_data.keys())
self.obj_indices = np.arange(len(self.image_names))
self.batch_size = batch_size
self.shuffle = shuffle
self.augment = augment
self.debug = debug
self.lock = threading.Lock() # mutex for input path
self.yield_lock = threading.Lock() # mutex for generator yielding of batch
self.init_count = 0
self.objects_id_generator = threadsafe_iter(get_objects_i_generator(len(self.image_names)))
self.crops_per_snapshot = crops_per_snapshot
self.crop_size = crop_size
self.cache = {}
if self.augment:
self.seq = iaa.Sequential([iaa.Fliplr(0.5),
iaa.Flipud(0.5),
iaa.GaussianBlur(sigma=(0, 5)),
iaa.Affine(rotate=(-180, 180),
translate_percent={'x': (-0.1, 0.1), 'y': (-0.1, 0.1)},
shear={'x': (-15, 15), 'y': (-15, 15)})],
random_order=True)
self.crop = iaa.CropToFixedSize(self.crop_size[0], self.crop_size[1])
self.crop_augmentations = iaa.Sequential([iaa.Cutout(nb_iterations=(1, 3), size=0.3, squared=False, cval=0),
iaa.CoarseDropout((0.0, 0.05), size_percent=(0.02, 0.25))],
random_order=True)
else:
self.seq = iaa.Identity()
def img_pro(old_path, new_path):
img_list = os.listdir(old_path)
num = 1
for img in img_list:
old_img = old_path + img
img_save = new_path + img
old_img = imageio.imread(old_img)
aug = iaa.Cutout(nb_iterations=250, size=[0.03, 0.04], cval=0)
img_mask = aug.augment_image(old_img)
print("img number:", num, 'processing img_name:', img)
num += 1
imageio.imwrite(img_save, img_mask)
def augment_merged(image, N=100):
"""
Given An Image Create Augment N augmented copies of the image
Since we prefer lists or numpy arrays, it is suggested to use OpenCV for opening images
"""
images = np.array([image for _ in range(N)], dtype=np.uint8)
seq = iaa.Sequential(
[
#iaa.Fliplr(0.5), # horizontal flips
#iaa.Flipud(0.5),
#iaa.CropAndPad(percent=(-0.10, 0.10),
# pad_mode=["constant", "mean","maximum"],
# pad_cval=(0, 128)), # random crops
# Small gaussian blur with random sigma between 0 and 0.5.
# But we only blur about 50% of all images.
#iaa.Sometimes(0.25,
# iaa.GaussianBlur(sigma=(0, 0.5))),
#iaa.CoarseDropout((0.0, 0.15), size_percent=(0.02, 0.15)),
# Strengthen or weaken the contrast in each image.
iaa.ContrastNormalization((0.7, 1.50)),
# Add gaussian noise.
# For 50% of all images, we sample the noise once per pixel.
# For the other 50% of all images, we sample the noise per pixel AND
# channel. This can change the color (not only brightness) of the
# pixels.
iaa.AdditiveGaussianNoise(
loc=0, scale=(0.0, 0.05 * 255), per_channel=0.5),
#iaa.PiecewiseAffine(scale=(0.01, 0.07)),
# Make some images brighter and some darker.
# In 20% of all cases, we sample the multiplier once per channel,
# which can end up changing the color of the images.
iaa.Multiply((0.75, 1.25), per_channel=.35),
#iaa.ChannelShuffle(0.5),
iaa.Cutout(nb_iterations=(1, 3),
size=(0.15, 0.25),
squared=False,
fill_mode="constant",
cval=(0, 255),
fill_per_channel=0.5)
# Apply affine transformations to each image.
# Scale/zoom them, translate/move them, rotate them and shear them.
#iaa.Affine(
# scale={"x": (0.75, 1.5), "y": (0.75, 1.5)},
# translate_percent={"x": (-0.2, 0.2), "y": (-0.2, 0.2)},
# rotate=(-20, 20),
# shear=(-8, 8)
#)
],
random_order=True) # apply augmenters in random order
images_aug = seq(images=images)
return images_aug
def post_composition_augmentation(image_array):
aug = iaa.Cutout(fill_mode="constant",
cval=(0, 255),
fill_per_channel=0.5,
nb_iterations=(1, 5),
size=0.25,
squared=False)
image_array = aug(image=image_array)
enhancer = ImageEnhance.Brightness(Image.fromarray(image_array))
return enhancer.enhance(random.uniform(0.3, 1))
def __init__(self,
thresh: float,
n: int = 3,
size: float = 0.3,
squared: bool = False):
self.name = "cutout"
assert 0.0 <= thresh <= 1.0
assert 0.0 <= size <= 1.0
self.thresh = thresh
self.transform = iaa.Cutout(nb_iterations=n,
size=size,
squared=squared,
cval=0)
def init_augmenter(img_mode="color"):
"""Initializes the augmenters used in the training dataset
:param config: the config object that contains all the
"""
ia.seed(10)
if img_mode == 'color':
return iaa.Sequential([
sometimes(iaa.Fliplr()),
iaa.MultiplyBrightness((0.6, 1.4)),
# TODO: try no ChangeColor or Brightness
sometimes(iaa.ChangeColorTemperature((5000, 7000))),
iaa.Crop(percent=(
(0, 0.50),
(0, 0.50),
(0, 0.50),
(0, 0.50)
)),
sometimes(iaa.OneOf([
iaa.Cutout(nb_iterations=(1, 4), size=0.2,
squared=False, cval=(0, 255), fill_mode="constant"),
iaa.Cutout(nb_iterations=(1, 4), size=0.2, squared=False, cval=(
0, 255), fill_mode="gaussian", fill_per_channel=True),
iaa.AdditiveGaussianNoise(scale=(0, 0.1*255))
]))
])
else:
return iaa.Sequential([
sometimes(iaa.Fliplr()),
iaa.Crop(percent=(
(0, 0.40),
(0, 0.40),
(0, 0.40),
(0, 0.40)
))
])
def __init__(self, augmentation_ratio=1.0,
flip_probability=0.5,
contrast_probability=0.2,
defocus_probability=0.1,
motion_blur_probability=0.2,
cutout_probability=0.2,
scale_probability=0.1,
shift_probability=0.2,
fog_probability=0.1):
rot = iaa.Rot90(ia.ALL)
shift = iaa.Affine(translate_percent=(-0.2, 0.2))
flip = iaa.OneOf([iaa.Fliplr(1), iaa.Flipud(1)])
scale = iaa.Affine(scale=(0.75, 1.25))
motion_blur = iaa.MotionBlur()
cutout = iaa.Cutout(
nb_iterations=(1, 3),
size=(0.01, 0.1),
squared=False)
fog = iaa.imgcorruptlike.Fog(severity=1)
defocus = iaa.imgcorruptlike.DefocusBlur(severity=1)
contrast = iaa.imgcorruptlike.Contrast(severity=1)
aug_weather = iaa.Sequential([
iaa.Sometimes(augmentation_ratio * fog_probability, fog),
], random_order=True)
aug_initial = iaa.Sequential([
rot,
iaa.Sometimes(flip_probability, flip),
iaa.Sometimes(augmentation_ratio * contrast_probability, contrast)
], random_order=True)
aug_camera = iaa.Sequential([
iaa.Sometimes(augmentation_ratio * defocus_probability, defocus),
iaa.Sometimes(augmentation_ratio *
motion_blur_probability, motion_blur),
], random_order=False)
aug_obstacles = iaa.Sequential([
iaa.Sometimes(augmentation_ratio * cutout_probability, cutout),
iaa.Sometimes(augmentation_ratio * scale_probability, scale),
iaa.Sometimes(shift_probability, shift)
], random_order=False)
self.augmentation_pipeline = iaa.Sequential(
[aug_weather, aug_initial, aug_camera, aug_obstacles],
random_order=False)
def img_aug(image, method):
# image = mpimg.imread(
# Path(config.base_dir, 'datasets', 'images', img_path))
# print('original image')
# ia.imshow(image)
# sequential
if method == 'seq':
aug = iaa.Sequential([
iaa.CropAndPad(percent=(-0.2, 0.2), pad_mode="edge"),
iaa.AddToHueAndSaturation((-60, 60)),
iaa.ElasticTransformation(alpha=90, sigma=9),
iaa.Cutout(),
],
random_order=True)
image = aug(image=image)
# print("water-like")
# ia.imshow(image)
# blend alpha
elif method == 'ba':
aug = iaa.BlendAlphaVerticalLinearGradient(iaa.AveragePooling(11),
start_at=(0.0, 1.0),
end_at=(0.0, 1.0))
image = aug(image=image)
# print('blend alpha')
# ia.imshow(image)
# mosaic
elif method == 'ba_box':
aug = iaa.BlendAlphaRegularGrid(nb_rows=2,
nb_cols=2,
foreground=iaa.Multiply(0.0),
background=iaa.AveragePooling(8),
alpha=[0.0, 0.0, 1.0])
image = aug(image=image)
# print('blend_alpha with boxes')
# ia.imshow(image)
return image
def __call__(self, sample):
img, annot = sample['img'], sample['annot']
bbs = BoundingBoxesOnImage([
BoundingBox(x1=ann[0],
y1=ann[1],
x2=ann[2],
y2=ann[3],
label=str(int(ann[4]))) for ann in annot
],
shape=img.shape)
aug = iaa.Cutout(nb_iterations=int(round(self.v)),
size=0.05,
fill_mode="gaussian")
img_aug, bbs_aug = aug(image=img, bounding_boxes=bbs)
annot_aug = np.array(
[[bb.x1, bb.y1, bb.x2, bb.y2,
np.float32(bb.label)] for bb in bbs_aug])
# the shape has to be at least (0,5)
if len(annot_aug) == 0:
annot_aug = np.zeros((0, 5))
return {'img': img_aug, 'annot': annot_aug}
def getImageAug():
seq = iaa.Sequential(
[
iaa.SomeOf(
(0, 2),
[
iaa.Identity(),
iaa.AverageBlur(k=((3, 5), (5, 7))),
iaa.Rotate((-45, 45)),
iaa.Affine(scale=(0.5, 0.95)),
iaa.Multiply((0.50, 1.1))
#,iaa.BlendAlphaRegularGrid(nb_rows=(4, 6), nb_cols=(1, 4),
# foreground=iaa.Multiply(0.0))
#,iaa.Cartoon()
,
iaa.Cutout(
nb_iterations=(1, 3), size=0.2, squared=False, cval=0),
iaa.Affine(shear=(-48, 48)),
iaa.Affine(translate_px={
"x": (-42, 42),
"y": (-36, 36)
}),
iaa.KeepSizeByResize(
iaa.Resize({
"height": (0.70, 0.90),
"width": (0.70, 0.90)
})),
iaa.CropAndPad(percent=(-0.2, 0.2))
#,iaa.PiecewiseAffine(scale=(0.01, 0.05))
,
iaa.PerspectiveTransform(scale=(0.01, 0.1))
#,iaa.WithPolarWarping(iaa.CropAndPad(percent=(-0.1, 0.1)))
#,iaa.ElasticTransformation(alpha=(0, 3.0), sigma=0.5)
])
#,iaa.SaveDebugImageEveryNBatches(folder_path, 100)
],
random_order=True)
return seq
def augment1(txt_path, save_txt_path):
"""
Augment data from the txt file.
"""
seq = iaa.Sequential(
[iaa.ElasticTransformation(alpha=90, sigma=9),
iaa.Cutout()],
random_order=False)
with open(txt_path, 'r') as fp:
lines = fp.readlines()
new_lines = []
for line in tqdm.tqdm(lines):
new_line = ""
cat_id, img_path = line.rstrip().split(' ')
new_line += cat_id
image = imageio.imread(img_path)
images_aug = seq(images=[image])[0]
img_path = img_path.replace('cloudy', 'augment').replace(
'dusky', 'augment').replace('foggy',
'augment').replace('sunny', 'augment')
new_line += ' ' + img_path + '\n'
new_lines.append(new_line)
images_aug = Image.fromarray(images_aug)
images_aug.save(img_path)
with open(save_txt_path, 'a+') as fp:
fp.writelines(new_lines)
# augment1("data/weather/train.txt", "data/weather/augment.txt")
# get_chips_augment(txt_path="data/weather/test.txt", save_name="chips_test", save_txt_path="data/weather/chips_test.txt")
# create_gif("C:/Users/18917/Documents/Python Scripts/pytorch/Lab/Pix2Pix-forlab/data-blur/6")
def cutout_fraction(magnitude: int, **kwargs) -> iaa.Cutout:
"""
Applies cutout to the image according to bbox information. This will
apply only to a single bounding box in the image. For the augmentation
to apply the policy correctly the image height and width along with the
bounding box height and width are required as keyword arguments.
Tensorflow Policy Equivalent: bbox_cutout
The cutout size is determined as a fraction of the bounding box size.
The cutout value in the policies is at a pixel level. The imgaug cutout
augmentation method requires the cutout to be a percentage of the image.
Passing the image height and width as kwargs will scale the cutout to the
appropriate percentage. Otherwise the imgaug default of 20% will be used.
Note: the cutout may not always be present in the bounding box dut to
randomness in the location of the cutout centre
:type magnitude: int
:param magnitude: magnitude of cutout
:param kwargs:
height: height of the image as int
width: width of the image as int
height_bbox: height of the bounding box as int
width_bbox: width of the bounding box as int
:rtype: iaa.Cutout
:return: Method to apply cutout to bounding boxes
"""
level = (magnitude / _MAX_MAGNITUDE) * CUTOUT_MAX_PAD_FRACTION
cutout_args = {}
if all(i in kwargs
for i in ['height', 'width', 'height_bbox', 'width_bbox']):
size = tuple([(level * kwargs['height_bbox']) / kwargs['height'],
(level * kwargs['width_bbox']) / kwargs['width']])
cutout_args['size'] = size
return iaa.Cutout(**cutout_args)
def __init__(self, da):
assert check_data_aug(da), 'Parameters for custom data augmentation missing. Should have: {}'.format(
keys_data_aug)
aug_geometric = [iaa.Affine(scale=(da['scalem'], da['scaleM']),
translate_percent={'x': (-da['trans'], da['trans']),
'y': (-da['trans'], da['trans'])},
rotate=(-da['rot'], da['rot']),
shear=(-da['shear'], da['shear']),
cval=FILL_COLOR),
iaa.PerspectiveTransform(scale=(0, da['pers']),
cval=FILL_COLOR, keep_size=True)]
aug_camera = [iaa.GaussianBlur(sigma=(0, da['sigma'])),
iaa.MotionBlur(k=(da['mot_km'], da['mot_kM']),
angle=(-da['mot_an'], da['mot_an']),
direction=(da['mot_dm'], da['mot_dM'])),
iaa.JpegCompression(compression=(da['jpegm'], da['jpegM'])),
iaa.LinearContrast(alpha=(da['con_alpham'], da['con_alphaM']),
per_channel=da['con_chan']),
iaa.MultiplyHueAndSaturation(mul=(da['col_mulm'], da['col_mulM']),
per_channel=da['col_chan'],
from_colorspace='BGR'),
iaa.AddToHueAndSaturation((da['col_addm'], da['col_addM']),
per_channel=da['col_chan'])
]
cutout = iaa.Cutout(nb_iterations=(0, da['co_num']),
size=(da['co_sm'], da['co_sM']),
squared=False,
cval=FILL_COLOR)
# Create a mix of all others
self.augmenter = iaa.Sequential([iaa.SomeOf((0, 1), aug_geometric), # none or 1
iaa.SomeOf((0, len(aug_camera) - 2), aug_camera), # from none to all-2
cutout], random_order=True) # mix the apply order