def test_StandardTransformer(self):
config = {
'raw': [{
'name': 'Normalize'
}, {
'name': 'RandomContrast',
'execution_probability': 0.5
}, {
'name': 'RandomFlip'
}, {
'name': 'RandomRotate90'
}, {
'name': 'ToTensor',
'expand_dims': True
}],
'label': [{
'name': 'RandomFlip'
}, {
'name': 'RandomRotate90'
}, {
'name': 'ToTensor',
'expand_dims': False,
'dtype': 'long'
}]
}
base_config = {'mean': 0, 'std': 1}
transformer = Transformer(config, base_config)
raw_transforms = transformer.raw_transform().transforms
assert raw_transforms[0].mean == 0
assert raw_transforms[0].std == 1
assert raw_transforms[1].execution_probability == 0.5
assert raw_transforms[4].expand_dims
label_transforms = transformer.label_transform().transforms
assert len(label_transforms) == 3
def test_BaseTransformer(self):
config = {
'raw': [{
'name': 'Normalize'
}, {
'name': 'ToTensor',
'expand_dims': True
}],
'label': [{
'name': 'ToTensor',
'expand_dims': False,
'dtype': 'long'
}],
'weight': [{
'name': 'ToTensor',
'expand_dims': False
}]
}
base_config = {'mean': 0, 'std': 1}
transformer = Transformer(config, base_config)
raw_transforms = transformer.raw_transform().transforms
assert raw_transforms[0].mean == 0
assert raw_transforms[0].std == 1
assert raw_transforms[1].expand_dims
label_transforms = transformer.label_transform().transforms
assert not label_transforms[0].expand_dims
assert label_transforms[0].dtype == 'long'
weight_transforms = transformer.weight_transform().transforms
assert not weight_transforms[0].expand_dims
def test_AnisotropicRotationTransformer(self):
config = {
'raw': [
{'name': 'Normalize'},
{'name': 'RandomContrast', 'execution_probability': 0.5},
{'name': 'RandomFlip'},
{'name': 'RandomRotate90'},
{'name': 'RandomRotate', 'angle_spectrum': 17, 'axes': [[2, 1]]},
{'name': 'ToTensor', 'expand_dims': True}
],
'label': [
{'name': 'RandomFlip'},
{'name': 'RandomRotate90'},
{'name': 'RandomRotate', 'angle_spectrum': 17, 'axes': [[2, 1]]},
{'name': 'ToTensor', 'expand_dims': False, 'dtype': 'long'}
]
}
transformer = Transformer(config, 0, 1)
raw_transforms = transformer.raw_transform().transforms
assert raw_transforms[0].mean == 0
assert raw_transforms[0].std == 1
assert raw_transforms[1].execution_probability == 0.5
assert raw_transforms[4].angle_spectrum == 17
assert raw_transforms[4].axes == [[2, 1]]
label_transforms = transformer.label_transform().transforms
assert len(label_transforms) == 4
def test_LabelToBoundaryTransformer(self):
config = {
'raw': [
{'name': 'Normalize'},
{'name': 'RandomContrast', 'execution_probability': 0.5},
{'name': 'RandomFlip'},
{'name': 'RandomRotate90'},
{'name': 'RandomRotate', 'angle_spectrum': 17, 'axes': [[2, 1]], 'mode': 'reflect'},
{'name': 'ToTensor', 'expand_dims': True}
],
'label': [
{'name': 'RandomFlip'},
{'name': 'RandomRotate90'},
{'name': 'RandomRotate', 'angle_spectrum': 17, 'axes': [[2, 1]], 'mode': 'reflect'},
{'name': 'LabelToBoundary', 'offsets': [2, 4, 6, 8]},
{'name': 'ToTensor', 'expand_dims': False, 'dtype': 'long'}
]
}
transformer = Transformer(config, 0, 1)
raw_transforms = transformer.raw_transform().transforms
assert raw_transforms[0].mean == 0
assert raw_transforms[0].std == 1
assert raw_transforms[1].execution_probability == 0.5
assert raw_transforms[4].angle_spectrum == 17
assert raw_transforms[4].axes == [[2, 1]]
assert raw_transforms[4].mode == 'reflect'
label_transforms = transformer.label_transform().transforms
assert label_transforms[2].angle_spectrum == 17
assert label_transforms[2].axes == [[2, 1]]
assert label_transforms[2].mode == 'reflect'
# 3 conv kernels per offset
assert len(label_transforms[3].kernels) == 12
def test_RandomLabelToBoundaryTransformer(self):
config = {
'raw': [{
'name': 'Normalize'
}, {
'name': 'RandomContrast',
'execution_probability': 0.5
}, {
'name': 'RandomFlip'
}, {
'name': 'RandomRotate90'
}, {
'name': 'RandomRotate',
'angle_spectrum': 17,
'axes': [[2, 1]],
'mode': 'reflect'
}, {
'name': 'ToTensor',
'expand_dims': True
}],
'label': [{
'name': 'RandomFlip'
}, {
'name': 'RandomRotate90'
}, {
'name': 'RandomRotate',
'angle_spectrum': 17,
'axes': [[2, 1]],
'mode': 'reflect'
}, {
'name': 'RandomLabelToAffinities',
'max_offset': 4
}, {
'name': 'ToTensor',
'expand_dims': False,
'dtype': 'long'
}]
}
base_config = {'mean': 0, 'std': 1}
transformer = Transformer(config, base_config)
label_transforms = transformer.label_transform().transforms
assert label_transforms[3].offsets == (1, 2, 3, 4)