def load_image(image_path):
image = read_image(image_path, format="BGR")
h = image.shape[0]
w = image.shape[1]
img_trans = TransformList(
[ResizeTransform(h=h, w=w, new_h=224, new_w=224)])
image = torch.tensor(img_trans.apply_image(image).copy()).permute(
2, 0, 1) # copy to make it writeable
return image, (w, h)
def get_transform(self, img):
h, w = img.shape[:2]
if min(h, w) >= self.short_edge_length[0]:
return NoOpTransform()
scale = self.short_edge_length[0] * 1.0 / min(h, w)
newh = h * scale
neww = w * scale
neww = int(neww + 0.5)
newh = int(newh + 0.5)
return ResizeTransform(h, w, newh, neww, self.interp)
def __call__(self, dataset_dict):
"""
Args:
dict: a dict in standard model input format. See tutorials for details.
Returns:
list[dict]:
a list of dicts, which contain augmented version of the input image.
The total number of dicts is ``len(min_sizes) * (2 if flip else 1)``.
Each dict has field "transforms" which is a TransformList,
containing the transforms that are used to generate this image.
"""
numpy_image = dataset_dict["image"].permute(1, 2, 0).numpy()
shape = numpy_image.shape
orig_shape = (dataset_dict["height"], dataset_dict["width"])
if shape[:2] != orig_shape:
# It transforms the "original" image in the dataset to the input image
pre_tfm = ResizeTransform(orig_shape[0], orig_shape[1], shape[0],
shape[1])
else:
pre_tfm = NoOpTransform()
# Create all combinations of augmentations to use
aug_candidates = [] # each element is a list[Augmentation]
for min_size in self.min_sizes:
resize = ResizeShortestEdge(min_size, self.max_size)
aug_candidates.append([resize]) # resize only
if self.flip:
flip = RandomFlip(prob=1.0)
aug_candidates.append([resize, flip]) # resize + flip
# Apply all the augmentations
ret = []
for aug in aug_candidates:
new_image, tfms = apply_augmentations(aug, np.copy(numpy_image))
torch_image = torch.from_numpy(
np.ascontiguousarray(new_image.transpose(2, 0, 1)))
dic = copy.deepcopy(dataset_dict)
dic["transforms"] = pre_tfm + tfms
dic["image"] = torch_image
if self.proposal_topk is not None:
image_shape = new_image.shape[:2] # h, w
transform_proposals(dic,
image_shape,
tfms,
proposal_topk=self.proposal_topk)
ret.append(dic)
return ret
def __call__(self, dataset_dict):
"""
Args:
dict: a detection dataset dict in standard format
Returns:
list[dict]:
a list of dataset dicts, which contain augmented version of the input image.
The total number of dicts is ``len(min_sizes) * (2 if flip else 1)``.
Each dict has field "transforms" which is a TransformList,
containing the transforms that are used to generate this image.
"""
numpy_image = dataset_dict["image"].permute(1, 2, 0).numpy()
shape = numpy_image.shape
orig_shape = (dataset_dict["height"], dataset_dict["width"])
if shape[:2] != orig_shape:
# It transforms the "original" image in the dataset to the input image
pre_tfm = ResizeTransform(orig_shape[0], orig_shape[1], shape[0],
shape[1])
else:
pre_tfm = NoOpTransform()
# Create all combinations of augmentations to use
tfm_gen_candidates = [] # each element is a list[TransformGen]
for min_size in self.min_sizes:
resize = ResizeShortestEdge(min_size, self.max_size)
tfm_gen_candidates.append([resize]) # resize only
if self.flip:
flip = RandomFlip(prob=1.0)
tfm_gen_candidates.append([resize, flip]) # resize + flip
if self.vertical_flip:
flip = RandomFlip(prob=1.0, horizontal=False, vertical=True)
tfm_gen_candidates.append([resize, flip]) # resize + flip
tfm_gen_candidates.append([resize, Rotation90Gen(prob=1.0)])
# tfm_gen_candidates.append([resize, CLAHEGen(prob=1.0, img_format=self.image_format)])
tfm_gen_candidates.append([
resize,
Rotation90Gen(prob=1.0),
Rotation90Gen(prob=1.0),
Rotation90Gen(prob=1.0)
])
# Apply all the augmentations
ret = []
for tfm_gen in tfm_gen_candidates:
new_image, tfms = apply_transform_gens(tfm_gen,
np.copy(numpy_image))
torch_image = torch.from_numpy(
np.ascontiguousarray(new_image.transpose(2, 0, 1)))
dic = copy.deepcopy(dataset_dict)
dic["transforms"] = pre_tfm + tfms
dic["image"] = torch_image
# pop original image width and height to prevent rescale boxes in postprocess of detector
dic.pop("width")
dic.pop("height")
ret.append(dic)
return ret
