|
|
import pytest |
|
|
import torch |
|
|
from torch.autograd import gradcheck |
|
|
|
|
|
import kornia |
|
|
from kornia.testing import assert_close, BaseTester |
|
|
|
|
|
|
|
|
class TestRgbToXyz(BaseTester): |
|
|
def test_smoke(self, device, dtype): |
|
|
C, H, W = 3, 4, 5 |
|
|
img = torch.rand(C, H, W, device=device, dtype=dtype) |
|
|
assert isinstance(kornia.color.rgb_to_xyz(img), torch.Tensor) |
|
|
|
|
|
@pytest.mark.parametrize("shape", [(1, 3, 4, 4), (2, 3, 2, 4), (3, 3, 4, 1), (3, 2, 1)]) |
|
|
def test_cardinality(self, device, dtype, shape): |
|
|
img = torch.ones(shape, device=device, dtype=dtype) |
|
|
assert kornia.color.rgb_to_xyz(img).shape == shape |
|
|
|
|
|
def test_exception(self, device, dtype): |
|
|
with pytest.raises(TypeError): |
|
|
assert kornia.color.rgb_to_xyz([0.0]) |
|
|
|
|
|
with pytest.raises(ValueError): |
|
|
img = torch.ones(1, 1, device=device, dtype=dtype) |
|
|
assert kornia.color.rgb_to_xyz(img) |
|
|
|
|
|
with pytest.raises(ValueError): |
|
|
img = torch.ones(2, 1, 1, device=device, dtype=dtype) |
|
|
assert kornia.color.rgb_to_xyz(img) |
|
|
|
|
|
def test_unit(self, device, dtype): |
|
|
data = torch.tensor( |
|
|
[ |
|
|
[ |
|
|
[ |
|
|
[0.9637, 0.0586, 0.6470, 0.6212, 0.9622], |
|
|
[0.8293, 0.4858, 0.8953, 0.2607, 0.3250], |
|
|
[0.5314, 0.4189, 0.8388, 0.8065, 0.2211], |
|
|
[0.9682, 0.2928, 0.4118, 0.2533, 0.0455], |
|
|
], |
|
|
[ |
|
|
[0.6936, 0.3457, 0.9466, 0.9937, 0.2692], |
|
|
[0.7485, 0.7320, 0.8323, 0.6889, 0.4831], |
|
|
[0.1865, 0.7439, 0.1366, 0.8858, 0.2077], |
|
|
[0.6227, 0.6140, 0.3936, 0.5024, 0.4157], |
|
|
], |
|
|
[ |
|
|
[0.6477, 0.9269, 0.7531, 0.7349, 0.9485], |
|
|
[0.4264, 0.8539, 0.9830, 0.2269, 0.1138], |
|
|
[0.3988, 0.1605, 0.6220, 0.0546, 0.1106], |
|
|
[0.2128, 0.5673, 0.0781, 0.1431, 0.3310], |
|
|
], |
|
|
] |
|
|
], |
|
|
device=device, |
|
|
dtype=dtype, |
|
|
) |
|
|
|
|
|
|
|
|
expected = torch.tensor( |
|
|
[ |
|
|
[ |
|
|
[ |
|
|
[0.7623584, 0.31501925, 0.7412189, 0.7441359, 0.66425407], |
|
|
[0.6866283, 0.61618143, 0.84423876, 0.39480132, 0.32732624], |
|
|
[0.3578189, 0.4677382, 0.50703406, 0.6592388, 0.18541752], |
|
|
[0.6603961, 0.44267434, 0.32468265, 0.30994105, 0.22713262], |
|
|
], |
|
|
[ |
|
|
[0.7477299, 0.32658678, 0.86891913, 0.89580274, 0.4656054], |
|
|
[0.7424382, 0.6884378, 0.8565741, 0.5644922, 0.4228247], |
|
|
[0.2751717, 0.63267857, 0.3209684, 0.8089483, 0.20354219], |
|
|
[0.6665957, 0.5423198, 0.37470126, 0.42349333, 0.33085644], |
|
|
], |
|
|
[ |
|
|
[0.7167665, 0.92310345, 0.8409531, 0.82877415, 0.95198023], |
|
|
[0.51042646, 0.9080406, 1.0505873, 0.30275896, 0.17200153], |
|
|
[0.4114541, 0.24927814, 0.62354034, 0.17305644, 0.13412625], |
|
|
[0.29514894, 0.6179093, 0.12908883, 0.20075734, 0.3649534], |
|
|
], |
|
|
] |
|
|
], |
|
|
device=device, |
|
|
dtype=dtype, |
|
|
) |
|
|
|
|
|
assert_close(kornia.color.rgb_to_xyz(data), expected) |
|
|
|
|
|
def test_forth_and_back(self, device, dtype): |
|
|
data = torch.rand(3, 4, 5, device=device, dtype=dtype) |
|
|
xyz = kornia.color.rgb_to_xyz |
|
|
rgb = kornia.color.xyz_to_rgb |
|
|
|
|
|
data_out = xyz(rgb(data)) |
|
|
assert_close(data_out, data) |
|
|
|
|
|
@pytest.mark.grad |
|
|
def test_gradcheck(self, device, dtype): |
|
|
B, C, H, W = 2, 3, 4, 4 |
|
|
img = torch.rand(B, C, H, W, device=device, dtype=torch.float64, requires_grad=True) |
|
|
assert gradcheck(kornia.color.rgb_to_xyz, (img,), raise_exception=True) |
|
|
|
|
|
@pytest.mark.jit |
|
|
def test_jit(self, device, dtype): |
|
|
B, C, H, W = 2, 3, 4, 4 |
|
|
img = torch.ones(B, C, H, W, device=device, dtype=dtype) |
|
|
op = kornia.color.rgb_to_xyz |
|
|
op_jit = torch.jit.script(op) |
|
|
assert_close(op(img), op_jit(img)) |
|
|
|
|
|
@pytest.mark.nn |
|
|
def test_module(self, device, dtype): |
|
|
B, C, H, W = 2, 3, 4, 4 |
|
|
img = torch.ones(B, C, H, W, device=device, dtype=dtype) |
|
|
ops = kornia.color.RgbToXyz().to(device, dtype) |
|
|
fcn = kornia.color.rgb_to_xyz |
|
|
assert_close(ops(img), fcn(img)) |
|
|
|
|
|
|
|
|
class TestXyzToRgb(BaseTester): |
|
|
def test_smoke(self, device, dtype): |
|
|
C, H, W = 3, 4, 5 |
|
|
img = torch.rand(C, H, W, device=device, dtype=dtype) |
|
|
assert isinstance(kornia.color.xyz_to_rgb(img), torch.Tensor) |
|
|
|
|
|
@pytest.mark.parametrize("shape", [(1, 3, 4, 4), (2, 3, 2, 4), (3, 3, 4, 1), (3, 2, 1)]) |
|
|
def test_cardinality(self, device, dtype, shape): |
|
|
img = torch.ones(shape, device=device, dtype=dtype) |
|
|
assert kornia.color.xyz_to_rgb(img).shape == shape |
|
|
|
|
|
def test_exception(self, device, dtype): |
|
|
with pytest.raises(TypeError): |
|
|
assert kornia.color.xyz_to_rgb([0.0]) |
|
|
|
|
|
with pytest.raises(ValueError): |
|
|
img = torch.ones(1, 1, device=device, dtype=dtype) |
|
|
assert kornia.color.xyz_to_rgb(img) |
|
|
|
|
|
with pytest.raises(ValueError): |
|
|
img = torch.ones(2, 1, 1, device=device, dtype=dtype) |
|
|
assert kornia.color.xyz_to_rgb(img) |
|
|
|
|
|
def test_unit(self, device, dtype): |
|
|
data = torch.tensor( |
|
|
[ |
|
|
[ |
|
|
[ |
|
|
[0.6315228, 0.4196397, 0.33123854, 0.3277169, 0.20501322], |
|
|
[0.44133404, 0.4823162, 0.528561, 0.51644784, 0.28001237], |
|
|
[0.31131047, 0.8453884, 0.4486181, 0.6015828, 0.42048606], |
|
|
[0.5472367, 0.48154795, 0.36668795, 0.39913517, 0.40271503], |
|
|
], |
|
|
[ |
|
|
[0.79137707, 0.501063, 0.3700857, 0.57410157, 0.15295872], |
|
|
[0.570678, 0.76664513, 0.48567873, 0.47680324, 0.2583247], |
|
|
[0.38080955, 0.9315215, 0.4404478, 0.50659215, 0.5984908], |
|
|
[0.5388581, 0.76993656, 0.4027568, 0.5952581, 0.68663263], |
|
|
], |
|
|
[ |
|
|
[0.86013114, 0.17629854, 0.83010703, 0.27881518, 0.30543375], |
|
|
[0.17009716, 0.61201245, 0.33521807, 0.15526368, 0.7401195], |
|
|
[0.34011865, 0.6541383, 0.96909684, 0.43090558, 0.70467836], |
|
|
[0.6738866, 0.47461915, 0.91508406, 0.44147202, 0.14099535], |
|
|
], |
|
|
] |
|
|
], |
|
|
device=device, |
|
|
dtype=dtype, |
|
|
) |
|
|
|
|
|
|
|
|
expected = torch.tensor( |
|
|
[ |
|
|
[ |
|
|
[ |
|
|
[0.4011656, 0.5017336, 0.09065688, 0.04048038, 0.2769511], |
|
|
[0.46811658, 0.07937729, 0.79911184, 0.8632158, 0.1413149], |
|
|
[0.2538725, 0.98146427, 0.29357457, 0.95588684, 0.09129933], |
|
|
[0.6090472, 0.14032376, 0.11294556, 0.15829873, 0.1792411], |
|
|
], |
|
|
[ |
|
|
[0.90825266, 0.54057753, 0.40771842, 0.7709542, 0.1009315], |
|
|
[0.64988965, 0.99616426, 0.41274834, 0.40036052, 0.24396756], |
|
|
[0.42678973, 0.95531154, 0.43172216, 0.3851813, 0.7444883], |
|
|
[0.5084846, 0.99737406, 0.4381809, 0.7481805, 0.9036418], |
|
|
], |
|
|
[ |
|
|
[0.78309417, 0.10751611, 0.82060075, 0.19588977, 0.3031369], |
|
|
[0.08796211, 0.51749885, 0.28474382, 0.09561294, 0.74540937], |
|
|
[0.2992335, 0.5486014, 0.95973116, 0.38571155, 0.64634556], |
|
|
[0.6330101, 0.3715171, 0.90575427, 0.36752605, 0.03138365], |
|
|
], |
|
|
] |
|
|
], |
|
|
device=device, |
|
|
dtype=dtype, |
|
|
) |
|
|
|
|
|
assert_close(kornia.color.xyz_to_rgb(data), expected, atol=1e-4, rtol=1e-4) |
|
|
|
|
|
def test_forth_and_back(self, device, dtype): |
|
|
data = torch.rand(3, 4, 5, device=device, dtype=dtype) |
|
|
xyz = kornia.color.rgb_to_xyz |
|
|
rgb = kornia.color.xyz_to_rgb |
|
|
|
|
|
data_out = rgb(xyz(data)) |
|
|
assert_close(data_out, data) |
|
|
|
|
|
@pytest.mark.grad |
|
|
def test_gradcheck(self, device, dtype): |
|
|
B, C, H, W = 2, 3, 4, 4 |
|
|
img = torch.rand(B, C, H, W, device=device, dtype=torch.float64, requires_grad=True) |
|
|
assert gradcheck(kornia.color.xyz_to_rgb, (img,), raise_exception=True) |
|
|
|
|
|
@pytest.mark.jit |
|
|
def test_jit(self, device, dtype): |
|
|
B, C, H, W = 2, 3, 4, 4 |
|
|
img = torch.ones(B, C, H, W, device=device, dtype=dtype) |
|
|
op = kornia.color.xyz_to_rgb |
|
|
op_jit = torch.jit.script(op) |
|
|
assert_close(op(img), op_jit(img)) |
|
|
|
|
|
@pytest.mark.nn |
|
|
def test_module(self, device, dtype): |
|
|
B, C, H, W = 2, 3, 4, 4 |
|
|
img = torch.ones(B, C, H, W, device=device, dtype=dtype) |
|
|
ops = kornia.color.XyzToRgb().to(device, dtype) |
|
|
fcn = kornia.color.xyz_to_rgb |
|
|
assert_close(ops(img), fcn(img)) |
|
|
|
