def test_convert_points_from_homogeneous(self):
# generate input data
batch_size = 2
points_h = torch.rand(batch_size, 2, 3)
points_h[..., -1] = 1.0
# to euclidean
points = tgm.convert_points_from_homogeneous(points_h)
error = utils.compute_mse(points_h[..., :2], points)
self.assertAlmostEqual(error.item(), 0.0, places=4)
# functional
self.assertTrue(
torch.allclose(points,
tgm.ConvertPointsFromHomogeneous()(points_h)))
def test_convert_points_batch(self, device_type):
# generate input data
points_h = torch.FloatTensor([[
[2, 1, 0],
], [
[0, 1, 2],
]]).to(torch.device(device_type))
expected = torch.FloatTensor([[
[2e6, 1e6],
], [
[0, 0.5],
]]).to(torch.device(device_type))
# to euclidean
points = tgm.convert_points_from_homogeneous(points_h)
assert_allclose(points, expected)
def test_convert_points_from_homogeneous(batch_shape, device_type):
# generate input data
points_h = torch.rand(batch_shape)
points_h = points_h.to(torch.device(device_type))
points_h[..., -1] = 1.0
# to euclidean
points = tgm.convert_points_from_homogeneous(points_h)
error = utils.compute_mse(points_h[..., :2], points)
assert pytest.approx(error.item(), 0.0)
# functional
assert torch.allclose(points, tgm.ConvertPointsFromHomogeneous()(points_h))
# evaluate function gradient
points = utils.tensor_to_gradcheck_var(points) # to var
assert gradcheck(tgm.convert_points_from_homogeneous, (points, ),
raise_exception=True)