π Geometric Computer Vision Library for Spatial AI
Python artificial-intelligencecomputer-visiondeep-learninghacktoberfestimage-processingmachine-learningneural-networkpythonpytorchroboticsspatial-ai
# LICENSE HEADER MANAGED BY add-license-header
#
# Copyright 2018 Kornia Team
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
import pytest
import torch
from kornia.enhance.integral import IntegralImage, IntegralTensor, integral_image, integral_tensor
from testing.base import BaseTester
class TestIntegralTensor(BaseTester):
def test_smoke(self, device, dtype):
tensor = torch.rand(1, 1, 4, 4, device=device, dtype=dtype)
output = integral_tensor(tensor)
assert output.shape == (1, 1, 4, 4)
@pytest.mark.parametrize("shape", [(1, 3, 4, 4), (2, 3, 2, 4), (3, 3, 4, 1)])
def test_cardinality(self, device, dtype, shape):
tensor = torch.rand(*shape, device=device, dtype=dtype)
output = integral_tensor(tensor)
assert output.shape == shape
def test_exception(self, device, dtype):
tensor = torch.rand(1, 1, 4, 4, device=device, dtype=dtype)
with pytest.raises(Exception):
integral_tensor(tensor, (0, 1, 2, 3, 4))
with pytest.raises(Exception):
integral_tensor(tensor, (4, 5))
with pytest.raises(Exception) as errinfo:
integral_tensor(tensor, ())
assert "dim must be a non-empty tuple." in str(errinfo)
@pytest.mark.parametrize("dim", [(1, 1), (-1, 1), (1, -1)])
def test_duplicate_dimensions(self, device, dtype, dim):
tensor = torch.ones(2, 2, device=device, dtype=dtype)
with pytest.raises(Exception, match="unique dimensions"):
integral_tensor(tensor, dim)
@pytest.mark.parametrize("dim", [(2,), (-3,), (0, -3)])
def test_out_of_range_dimensions(self, device, dtype, dim):
# -3 + 2 = -1 is a valid cumsum axis, so a missing bounds check would integrate the last axis silently.
tensor = torch.ones(2, 2, device=device, dtype=dtype)
with pytest.raises(Exception, match="valid dimensions"):
integral_tensor(tensor, dim)
@pytest.mark.parametrize("dim", [(0, 1), (1, 0), (-2, -1)])
def test_distinct_dimensions(self, device, dtype, dim):
tensor = torch.ones(2, 2, device=device, dtype=dtype)
output = integral_tensor(tensor, dim)
expected = torch.tensor([[1, 2], [2, 4]], device=device, dtype=dtype)
self.assert_close(output, expected)
def test_module(self, device, dtype):
mod = IntegralTensor()
op = integral_tensor
tensor = torch.rand(1, 1, 4, 4, device=device, dtype=dtype)
self.assert_close(mod(tensor), op(tensor))
def test_gradcheck(self, device):
tensor = torch.rand(1, 1, 4, 4, device=device, dtype=torch.float64, requires_grad=True)
self.gradcheck(integral_tensor, (tensor,))
def test_value(self, device, dtype):
tensor = torch.tensor([[[[1.0, 2.0, 3.0], [4.0, 5.0, 6.0], [7.0, 8.0, 9.0]]]], device=device, dtype=dtype)
output = integral_tensor(tensor)
expected = torch.tensor([[[[1.0, 3.0, 6.0], [4.0, 9.0, 15.0], [7.0, 15.0, 24.0]]]], device=device, dtype=dtype)
self.assert_close(output, expected)
class TestIntegralImage(BaseTester):
def test_smoke(self, device, dtype):
tensor = torch.rand(1, 1, 4, 4, device=device, dtype=dtype)
output = integral_image(tensor)
assert output.shape == (1, 1, 4, 4)
@pytest.mark.parametrize("shape", [(1, 3, 4, 4), (2, 3, 2, 4), (3, 3, 4, 1)])
def test_cardinality(self, device, dtype, shape):
tensor = torch.rand(*shape, device=device, dtype=dtype)
output = integral_image(tensor)
assert output.shape == shape
def test_exception(self, device, dtype):
tensor = torch.rand(4, device=device, dtype=dtype)
with pytest.raises(Exception):
integral_image(tensor)
def test_module(self, device, dtype):
mod = IntegralImage()
op = integral_image
tensor = torch.rand(1, 1, 4, 4, device=device, dtype=dtype)
self.assert_close(mod(tensor), op(tensor))
def test_gradcheck(self, device):
tensor = torch.rand(1, 1, 4, 4, device=device, dtype=torch.float64, requires_grad=True)
self.gradcheck(integral_image, (tensor,))
def test_values(self, device, dtype):
tensor = torch.tensor(
[[[[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12], [13, 14, 15, 16]]]], device=device, dtype=dtype
)
output = integral_image(tensor)
expected = torch.tensor(
[[[[1, 3, 6, 10], [6.0, 14.0, 24.0, 36.0], [15.0, 33.0, 54.0, 78.0], [28.0, 60.0, 96.0, 136.0]]]],
device=device,
dtype=dtype,
)
self.assert_close(output, expected)