Skip to content

🐍 Geometric Computer Vision Library for Spatial AI

test(geometry): fill the empty exception and cardinality stubs in the So3 and Se3 tests (#6292)

Adarsh Mishrac17c2e26deec
diff --git a/tests/geometry/liegroup/test_se3.py b/tests/geometry/liegroup/test_se3.py
index 829db2b1a..5113aec83 100644
--- a/tests/geometry/liegroup/test_se3.py
+++ b/tests/geometry/liegroup/test_se3.py
@@ -20,7 +20,7 @@ import math
import pytest
import torch
-from kornia.core.exceptions import DeviceError, TypeCheckError
+from kornia.core.exceptions import DeviceError, ShapeError, TypeCheckError
from kornia.geometry.conversions import euler_from_quaternion, rotation_matrix_to_quaternion
from kornia.geometry.liegroup import Se3, So3
from kornia.geometry.pose import NamedPose
@@ -79,9 +79,18 @@ class TestSe3(BaseTester):
se: Se3 = self._make_rand_se3d(device, dtype, batch_size)
assert se.r.q.shape[0] == batch_size
- # TODO: implement me
def test_exception(self, device, dtype):
- pass
+ with pytest.raises(ShapeError, match=r"Shape mismatch at dimension 1: expected 3, got 2"):
+ Se3(So3.identity(1, device, dtype), torch.zeros(1, 2, device=device, dtype=dtype))
+
+ with pytest.raises(ShapeError, match=r"Shape mismatch at dimension 1: expected 6, got 5"):
+ Se3.exp(torch.zeros(1, 5, device=device, dtype=dtype))
+
+ with pytest.raises(ShapeError, match=r"Shape mismatch at dimension 1: expected 4, got 3"):
+ Se3.from_matrix(torch.eye(3, device=device, dtype=dtype).repeat(1, 1, 1))
+
+ with pytest.raises(TypeError, match=r"Unsupported type"):
+ Se3(So3.identity(1, device, dtype), torch.zeros(1, 3, device=device, dtype=dtype)) * 4
def test_gradcheck(self, device):
v = torch.tensor([[1.0, 2.0, 3.0, 0.3, -0.4, 0.5]], device=device, dtype=torch.float64)
diff --git a/tests/geometry/liegroup/test_so2.py b/tests/geometry/liegroup/test_so2.py
index e44464aa3..90a1a2686 100644
--- a/tests/geometry/liegroup/test_so2.py
+++ b/tests/geometry/liegroup/test_so2.py
@@ -166,9 +166,10 @@ class TestSo2(BaseTester):
with pytest.raises(BaseError):
assert So2.identity(batch_size=0)
- # TODO: implement me
def test_gradcheck(self, device):
- pass
+ theta = torch.tensor([[0.4], [-0.3]], device=device, dtype=torch.float64)
+ self.gradcheck(lambda x: So2.exp(x).matrix(), (theta,))
+ self.gradcheck(lambda x: So2.exp(x).log(), (theta,))
# TODO: implement me
def test_jit(self, device, dtype):
diff --git a/tests/geometry/liegroup/test_so3.py b/tests/geometry/liegroup/test_so3.py
index 05d51ef80..e32f4935d 100644
--- a/tests/geometry/liegroup/test_so3.py
+++ b/tests/geometry/liegroup/test_so3.py
@@ -21,6 +21,7 @@ import pytest
import torch
from torch import nn
+from kornia.core.exceptions import ShapeError, TypeCheckError
from kornia.geometry.conversions import euler_from_quaternion
from kornia.geometry.liegroup import So3
from kornia.geometry.liegroup.so3 import _so3_small_angle_coefficients
@@ -42,13 +43,29 @@ class TestSo3(BaseTester):
assert isinstance(s, So3)
self.assert_close(s.q.data, q.data)
- # TODO: implement me
- def test_cardinality(self, device, dtype):
- pass
+ @pytest.mark.parametrize("batch_size", [1, 2, 5])
+ def test_cardinality(self, device, dtype, batch_size):
+ so: So3 = So3.random(batch_size, device, dtype)
+ assert so.q.shape[0] == batch_size
- # TODO: implement me
def test_exception(self, device, dtype):
- pass
+ with pytest.raises(ValueError, match=r"Quaternion input must have last dimension == 4"):
+ So3(Quaternion(torch.zeros(1, 3, device=device, dtype=dtype)))
+
+ with pytest.raises(TypeCheckError, match=r"Type mismatch: expected Quaternion"):
+ So3(torch.zeros(1, 4, device=device, dtype=dtype))
+
+ with pytest.raises(ShapeError, match=r"Shape mismatch at dimension 1: expected 3, got 2"):
+ So3.exp(torch.zeros(1, 2, device=device, dtype=dtype))
+
+ with pytest.raises(ShapeError, match=r"Shape mismatch at dimension 1: expected 4, got 3"):
+ So3.from_wxyz(torch.zeros(1, 3, device=device, dtype=dtype))
+
+ with pytest.raises(ValueError, match=r"Input size must be a \(\*, 3, 3\) tensor"):
+ So3.from_matrix(torch.zeros(3, device=device, dtype=dtype))
+
+ with pytest.raises(TypeError, match=r"Not So3 or torch.Tensor type"):
+ So3.identity(1, device, dtype) * [1.0, 2.0, 3.0]
def test_gradcheck(self, device):
v = torch.tensor([[0.3, -0.4, 0.5]], device=device, dtype=torch.float64)