GaussianBlur CV-CUDA Backend

test/test_transforms_v2.py

@@ -21,6 +21,7 @@
 import torchvision.transforms.v2 as transforms
 
 from common_utils import (
+    assert_close,
     assert_equal,
     cache,
     cpu_and_cuda,
@@ -41,7 +42,6 @@
 )
 
 from torch import nn
-from torch.testing import assert_close
 from torch.utils._pytree import tree_flatten, tree_map
 from torch.utils.data import DataLoader, default_collate
 from torchvision import tv_tensors
@@ -3936,7 +3936,15 @@ def test_kernel_video(self):
 
     @pytest.mark.parametrize(
         "make_input",
-        [make_image_tensor, make_image_pil, make_image, make_video],
+        [
+            make_image_tensor,
+            make_image_pil,
+            make_image,
+            make_video,
+            pytest.param(
+                make_image_cvcuda, marks=pytest.mark.skipif(not CVCUDA_AVAILABLE, reason="CVCUDA is not available")
+            ),
+        ],
     )
     def test_functional(self, make_input):
         check_functional(F.gaussian_blur, make_input(), kernel_size=(3, 3))
@@ -3948,14 +3956,31 @@ def test_functional(self, make_input):
             (F._misc._gaussian_blur_image_pil, PIL.Image.Image),
             (F.gaussian_blur_image, tv_tensors.Image),
             (F.gaussian_blur_video, tv_tensors.Video),
+            pytest.param(
+                F._misc._gaussian_blur_image_cvcuda,
+                None,
+                marks=pytest.mark.skipif(not CVCUDA_AVAILABLE, reason="CVCUDA is not available"),
+            ),
         ],
     )
     def test_functional_signature(self, kernel, input_type):
+        if kernel is F._misc._gaussian_blur_image_cvcuda:
+            input_type = _import_cvcuda().Tensor
         check_functional_kernel_signature_match(F.gaussian_blur, kernel=kernel, input_type=input_type)
 
     @pytest.mark.parametrize(
         "make_input",
-        [make_image_tensor, make_image_pil, make_image, make_bounding_boxes, make_segmentation_mask, make_video],
+        [
+            make_image_tensor,
+            make_image_pil,
+            make_image,
+            make_bounding_boxes,
+            make_segmentation_mask,
+            make_video,
+            pytest.param(
+                make_image_cvcuda, marks=pytest.mark.skipif(not CVCUDA_AVAILABLE, reason="CVCUDA is not available")
+            ),
+        ],
     )
     @pytest.mark.parametrize("device", cpu_and_cuda())
     @pytest.mark.parametrize("sigma", [5, 2.0, (0.5, 2), [1.3, 2.7]])
@@ -4018,11 +4043,22 @@ def test_make_params(self, sigma):
             ((1, 26, 28), (23, 23), 1.7),
         ],
     )
-    @pytest.mark.parametrize("dtype", [torch.float32, torch.float64, torch.float16])
+    @pytest.mark.parametrize("dtype", [torch.uint8, torch.float32, torch.float64, torch.float16])
     @pytest.mark.parametrize("device", cpu_and_cuda())
-    def test_functional_image_correctness(self, dimensions, kernel_size, sigma, dtype, device):
+    @pytest.mark.parametrize(
+        "input_type",
+        [
+            tv_tensors.Image,
+            pytest.param(
+                "cvcuda.Tensor", marks=pytest.mark.skipif(not CVCUDA_AVAILABLE, reason="CVCUDA not available")
+            ),
+        ],
+    )
+    def test_functional_image_correctness(self, dimensions, kernel_size, sigma, dtype, device, input_type):
         if dtype is torch.float16 and device == "cpu":
             pytest.skip("The CPU implementation of float16 on CPU differs from opencv")
+        if (dtype != torch.float32 and dtype != torch.uint8) and input_type == "cvcuda.Tensor":
+            pytest.skip("CVCUDA does not support non-float32 or uint8 dtypes for gaussian blur")
 
         num_channels, height, width = dimensions
 
@@ -4042,9 +4078,17 @@ def test_functional_image_correctness(self, dimensions, kernel_size, sigma, dtyp
             device=device,
         )
 
-        actual = F.gaussian_blur_image(image, kernel_size=kernel_size, sigma=sigma)
+        if input_type == "cvcuda.Tensor":
+            image = image.unsqueeze(0)
+            image = F.to_cvcuda_tensor(image)
 
-        torch.testing.assert_close(actual, expected, rtol=0, atol=1)
+        actual = F.gaussian_blur(image, kernel_size=kernel_size, sigma=sigma)
+
+        if input_type == "cvcuda.Tensor":
+            actual = F.cvcuda_to_tensor(actual)
+            actual = actual.squeeze(0).to(device=device)
+
+        assert_close(actual, expected, rtol=0, atol=1)
 
 
 class TestGaussianNoise:

torchvision/transforms/v2/_misc.py

@@ -9,6 +9,7 @@
 
 from torchvision import transforms as _transforms, tv_tensors
 from torchvision.transforms.v2 import functional as F, Transform
+from torchvision.transforms.v2.functional._utils import _is_cvcuda_available, _is_cvcuda_tensor
 
 from ._utils import (
     _parse_labels_getter,
@@ -20,6 +21,8 @@
     is_pure_tensor,
 )
 
+CVCUDA_AVAILABLE = _is_cvcuda_available()
+
 
 # TODO: do we want/need to expose this?
 class Identity(Transform):
@@ -192,6 +195,9 @@ class GaussianBlur(Transform):
 
     _v1_transform_cls = _transforms.GaussianBlur
 
+    if CVCUDA_AVAILABLE:
+        _transformed_types = Transform._transformed_types + (_is_cvcuda_tensor,)
+
     def __init__(
         self, kernel_size: Union[int, Sequence[int]], sigma: Union[int, float, Sequence[float]] = (0.1, 2.0)
     ) -> None:

torchvision/transforms/v2/_utils.py

@@ -16,7 +16,7 @@
 
 from torchvision.transforms.transforms import _check_sequence_input, _setup_angle, _setup_size  # noqa: F401
 from torchvision.transforms.v2.functional import get_dimensions, get_size, is_pure_tensor
-from torchvision.transforms.v2.functional._utils import _FillType, _FillTypeJIT
+from torchvision.transforms.v2.functional._utils import _FillType, _FillTypeJIT, _is_cvcuda_tensor
 
 
 def _setup_number_or_seq(arg: int | float | Sequence[int | float], name: str) -> Sequence[float]:
@@ -182,7 +182,7 @@ def query_chw(flat_inputs: list[Any]) -> tuple[int, int, int]:
     chws = {
         tuple(get_dimensions(inpt))
         for inpt in flat_inputs
-        if check_type(inpt, (is_pure_tensor, tv_tensors.Image, PIL.Image.Image, tv_tensors.Video))
+        if check_type(inpt, (is_pure_tensor, tv_tensors.Image, PIL.Image.Image, tv_tensors.Video, _is_cvcuda_tensor))
     }
     if not chws:
         raise TypeError("No image or video was found in the sample")
@@ -207,6 +207,7 @@ def query_size(flat_inputs: list[Any]) -> tuple[int, int]:
                 tv_tensors.Mask,
                 tv_tensors.BoundingBoxes,
                 tv_tensors.KeyPoints,
+                _is_cvcuda_tensor,
             ),
         )
     }

torchvision/transforms/v2/functional/_geometry.py

@@ -2,7 +2,7 @@
 import numbers
 import warnings
 from collections.abc import Sequence
-from typing import Any, Optional, TYPE_CHECKING, Union
+from typing import Any, Optional, Union, TYPE_CHECKING
 
 import PIL.Image
 import torch

torchvision/transforms/v2/functional/_meta.py

@@ -51,6 +51,16 @@ def get_dimensions_video(video: torch.Tensor) -> list[int]:
     return get_dimensions_image(video)
 
 
+def get_dimensions_image_cvcuda(image: "cvcuda.Tensor") -> list[int]:
+    # CV-CUDA tensor is always in NHWC layout
+    # get_dimensions is CHW
+    return [image.shape[3], image.shape[1], image.shape[2]]
+
+
+if CVCUDA_AVAILABLE:
+    _register_kernel_internal(get_dimensions, cvcuda.Tensor)(get_dimensions_image_cvcuda)
+
+
 def get_num_channels(inpt: torch.Tensor) -> int:
     if torch.jit.is_scripting():
         return get_num_channels_image(inpt)
@@ -87,6 +97,16 @@ def get_num_channels_video(video: torch.Tensor) -> int:
 get_image_num_channels = get_num_channels
 
 
+def get_num_channels_image_cvcuda(image: "cvcuda.Tensor") -> int:
+    # CV-CUDA tensor is always in NHWC layout
+    # get_num_channels is C
+    return image.shape[3]
+
+
+if CVCUDA_AVAILABLE:
+    _register_kernel_internal(get_num_channels, cvcuda.Tensor)(get_num_channels_image_cvcuda)
+
+
 def get_size(inpt: torch.Tensor) -> list[int]:
     if torch.jit.is_scripting():
         return get_size_image(inpt)
@@ -125,7 +145,7 @@ def get_size_image_cvcuda(image: "cvcuda.Tensor") -> list[int]:
 
 
 if CVCUDA_AVAILABLE:
-    _get_size_image_cvcuda = _register_kernel_internal(get_size, cvcuda.Tensor)(get_size_image_cvcuda)
+    _register_kernel_internal(get_size, _import_cvcuda().Tensor)(get_size_image_cvcuda)
 
 
 @_register_kernel_internal(get_size, tv_tensors.Video, tv_tensor_wrapper=False)

torchvision/transforms/v2/functional/_misc.py

@@ -1,5 +1,5 @@
 import math
-from typing import Optional
+from typing import List, Optional, Tuple, TYPE_CHECKING
 
 import PIL.Image
 import torch
@@ -13,7 +13,14 @@
 
 from ._meta import _convert_bounding_box_format
 
-from ._utils import _get_kernel, _register_kernel_internal, is_pure_tensor
+from ._utils import _get_kernel, _import_cvcuda, _is_cvcuda_available, _register_kernel_internal, is_pure_tensor
+
+CVCUDA_AVAILABLE = _is_cvcuda_available()
+
+if TYPE_CHECKING:
+    import cvcuda  # type: ignore[import-not-found]
+if CVCUDA_AVAILABLE:
+    cvcuda = _import_cvcuda()  # noqa: F811
 
 
 def normalize(
@@ -99,11 +106,10 @@ def _get_gaussian_kernel2d(
     return kernel2d
 
 
-@_register_kernel_internal(gaussian_blur, torch.Tensor)
-@_register_kernel_internal(gaussian_blur, tv_tensors.Image)
-def gaussian_blur_image(
-    image: torch.Tensor, kernel_size: list[int], sigma: Optional[list[float]] = None
-) -> torch.Tensor:
+def _validate_kernel_size_and_sigma(
+    kernel_size: List[int],
+    sigma: Optional[List[float]] = None,
+) -> Tuple[List[int], List[float]]:
     # TODO: consider deprecating integers from sigma on the future
     if isinstance(kernel_size, int):
         kernel_size = [kernel_size, kernel_size]
@@ -132,6 +138,16 @@ def gaussian_blur_image(
         if s <= 0.0:
             raise ValueError(f"sigma should have positive values. Got {sigma}")
 
+    return kernel_size, sigma
+
+
+@_register_kernel_internal(gaussian_blur, torch.Tensor)
+@_register_kernel_internal(gaussian_blur, tv_tensors.Image)
+def gaussian_blur_image(
+    image: torch.Tensor, kernel_size: list[int], sigma: Optional[list[float]] = None
+) -> torch.Tensor:
+    kernel_size, sigma = _validate_kernel_size_and_sigma(kernel_size, sigma)
+
     if image.numel() == 0:
         return image
 
@@ -181,6 +197,25 @@ def gaussian_blur_video(
     return gaussian_blur_image(video, kernel_size, sigma)
 
 
+def _gaussian_blur_image_cvcuda(
+    image: "cvcuda.Tensor", kernel_size: list[int], sigma: Optional[list[float]] = None
+) -> "cvcuda.Tensor":
+    cvcuda = _import_cvcuda()
+
+    kernel_size, sigma = _validate_kernel_size_and_sigma(kernel_size, sigma)
+
+    return cvcuda.gaussian(
+        image,
+        tuple(kernel_size),
+        tuple(sigma),
+        border=cvcuda.Border.REFLECT101,
+    )
+
+
+if CVCUDA_AVAILABLE:
+    _register_kernel_internal(gaussian_blur, _import_cvcuda().Tensor)(_gaussian_blur_image_cvcuda)
+
+
 def gaussian_noise(inpt: torch.Tensor, mean: float = 0.0, sigma: float = 0.1, clip: bool = True) -> torch.Tensor:
     """See :class:`~torchvision.transforms.v2.GaussianNoise`"""
     if torch.jit.is_scripting():