def test_torch(self, init_param, img: torch.Tensor, track_meta: bool,
device):
set_track_meta(track_meta)
img = img.to(device)
xform = Flip(init_param)
res = xform(img)
self.assertEqual(img.shape, res.shape)
if track_meta:
self.assertIsInstance(res, MetaTensor)
else:
self.assertNotIsInstance(res, MetaTensor)
self.assertIsInstance(res, torch.Tensor)
with self.assertRaisesRegex(ValueError, "MetaTensor"):
xform.inverse(res)
def test_correct_results(self, _, spatial_axis):
flip = Flip(spatial_axis=spatial_axis)
expected = list()
for channel in self.imt[0]:
expected.append(np.flip(channel, spatial_axis))
expected = np.stack(expected)
self.assertTrue(np.allclose(expected, flip(self.imt[0])))
def test_tranform_array(self, input):
transforms = Compose(
[Range("random flip")(Flip()),
Range()(ToTensor())])
# Apply transforms
output = transforms(input)
# Decorate with NVTX Range
transforms1 = Range()(transforms)
transforms2 = Range("Transforms2")(transforms)
transforms3 = Range(name="Transforms3", methods="__call__")(transforms)
# Apply transforms with Range
output1 = transforms1(input)
output2 = transforms2(input)
output3 = transforms3(input)
# Check the outputs
self.assertIsInstance(output, torch.Tensor)
self.assertIsInstance(output1, torch.Tensor)
self.assertIsInstance(output2, torch.Tensor)
self.assertIsInstance(output3, torch.Tensor)
np.testing.assert_equal(output.numpy(), output1.numpy())
np.testing.assert_equal(output.numpy(), output1.numpy())
np.testing.assert_equal(output.numpy(), output3.numpy())
def test_nvtx_transfroms_array(self, input):
# with prob == 0.0
transforms = Compose([
RandMark("Mark: Transforms Start!"),
RandRangePush("Range: RandFlip"),
RandFlip(prob=0.0),
RandRangePop(),
RangePush("Range: ToTensor"),
ToTensor(),
RangePop(),
Mark("Mark: Transforms End!"),
])
output = transforms(input)
self.assertIsInstance(output, torch.Tensor)
np.testing.assert_array_equal(input, output)
# with prob == 1.0
transforms = Compose([
RandMark("Mark: Transforms Start!"),
RandRangePush("Range: RandFlip"),
RandFlip(prob=1.0),
RandRangePop(),
RangePush("Range: ToTensor"),
ToTensor(),
RangePop(),
Mark("Mark: Transforms End!"),
])
output = transforms(input)
self.assertIsInstance(output, torch.Tensor)
np.testing.assert_array_equal(input, Flip()(output.numpy()))
def test_nvtx_transfroms_dict(self, input):
# with prob == 0.0
transforms = Compose([
RandMarkD("Mark: Transforms (p=0) Start!"),
RandRangePushD("Range: RandFlipD"),
RandFlipD(keys="image", prob=0.0),
RandRangePopD(),
RangePushD("Range: ToTensorD"),
ToTensorD(keys=("image")),
RangePopD(),
MarkD("Mark: Transforms (p=0) End!"),
])
output = transforms(input)
self.assertIsInstance(output["image"], torch.Tensor)
np.testing.assert_array_equal(input["image"], output["image"])
# with prob == 1.0
transforms = Compose([
RandMarkD("Mark: Transforms (p=1) Start!"),
RandRangePushD("Range: RandFlipD"),
RandFlipD(keys="image", prob=1.0),
RandRangePopD(),
RangePushD("Range: ToTensorD"),
ToTensorD(keys=("image")),
RangePopD(),
MarkD("Mark: Transforms (p=1) End!"),
])
output = transforms(input)
self.assertIsInstance(output["image"], torch.Tensor)
np.testing.assert_array_equal(input["image"],
Flip()(output["image"].numpy()))
def test_inverse_array(self, use_compose, dtype, device):
img: MetaTensor
tr = Compose([
AddChannel(),
Orientation("RAS"),
Flip(1),
Spacing([1.0, 1.2, 0.9], align_corners=False)
])
num_invertible = len(
[i for i in tr.transforms if isinstance(i, InvertibleTransform)])
# forward
img = tr(self.get_image(dtype, device))
self.assertEqual(len(img.applied_operations), num_invertible)
# inverse with Compose
if use_compose:
img = tr.inverse(img)
self.assertEqual(len(img.applied_operations), 0)
# inverse individually
else:
_tr: InvertibleTransform
num_to_inverse = num_invertible
for _tr in tr.transforms[::-1]:
if isinstance(_tr, InvertibleTransform):
img = _tr.inverse(img)
num_to_inverse -= 1
self.assertEqual(len(img.applied_operations),
num_to_inverse)
def test_correct_results(self, _, spatial_axis):
for p in TEST_NDARRAYS:
im = p(self.imt[0])
flip = Flip(spatial_axis=spatial_axis)
expected = [np.flip(channel, spatial_axis) for channel in self.imt[0]]
expected = np.stack(expected)
result = flip(im)
assert_allclose(result, p(expected))
def test_correct_results(self, _, spatial_axis):
for p in TEST_NDARRAYS_ALL:
im = p(self.imt[0])
flip = Flip(spatial_axis=spatial_axis)
expected = [
np.flip(channel, spatial_axis) for channel in self.imt[0]
]
expected = np.stack(expected)
result = flip(im)
assert_allclose(result, p(expected), type_test="tensor")
test_local_inversion(flip, result, im)
def test_tranform_randomized(self, input):
# Compose deterministic and randomized transforms
transforms = Compose([
Range("flip")(Flip()),
Rotate90(),
Range()(RandAdjustContrast(prob=0.0)),
Range("random flip")(RandFlip(prob=1.0)),
ToTensor(),
])
# Apply transforms
output = transforms(input)
# Decorate with NVTX Range
transforms1 = Range()(transforms)
transforms2 = Range("Transforms2")(transforms)
transforms3 = Range(name="Transforms3", methods="__call__")(transforms)
# Apply transforms with Range
output1 = transforms1(input)
output2 = transforms2(input)
output3 = transforms3(input)
# Check if the outputs are equal
self.assertIsInstance(output, torch.Tensor)
self.assertIsInstance(output1, torch.Tensor)
self.assertIsInstance(output2, torch.Tensor)
self.assertIsInstance(output3, torch.Tensor)
np.testing.assert_equal(output.numpy(), output1.numpy())
np.testing.assert_equal(output.numpy(), output2.numpy())
np.testing.assert_equal(output.numpy(), output3.numpy())
# Check if the first randomized is RandAdjustContrast
for tran in transforms.transforms:
if isinstance(tran, Randomizable):
self.assertIsInstance(tran, RandAdjustContrast)
break
def test_invalid_inputs(self, _, spatial_axis, raises):
with self.assertRaises(raises):
flip = Flip(spatial_axis)
flip(self.imt[0])
TEST_CASE_ARRAY_0 = [np.random.randn(3, 3)]
TEST_CASE_ARRAY_1 = [np.random.randn(3, 10, 10)]
TEST_CASE_DICT_0 = [{"image": np.random.randn(3, 3)}]
TEST_CASE_DICT_1 = [{"image": np.random.randn(3, 10, 10)}]
TEST_CASE_TORCH_0 = [torch.randn(3, 3)]
TEST_CASE_TORCH_1 = [torch.randn(3, 10, 10)]
TEST_CASE_WRAPPER = [np.random.randn(3, 10, 10)]
TEST_CASE_RECURSIVE_0 = [
torch.randn(3, 3),
Compose([
ToNumpy(),
Flip(),
RandAdjustContrast(prob=0.0),
RandFlip(prob=1.0),
ToTensor()
]),
]
TEST_CASE_RECURSIVE_1 = [
torch.randn(3, 3),
Compose([
ToNumpy(),
Flip(),
Compose([RandAdjustContrast(prob=0.0),
RandFlip(prob=1.0)]),
ToTensor()
]),
]
