simplify PIL comparisions

Author: justincdavis

test/common_utils.py

@@ -20,13 +20,15 @@
 from torch.testing._comparison import BooleanPair, NonePair, not_close_error_metas, NumberPair, TensorLikePair
 from torchvision import io, tv_tensors
 from torchvision.transforms._functional_tensor import _max_value as get_max_value
-from torchvision.transforms.v2.functional import to_cvcuda_tensor, to_image, to_pil_image
+from torchvision.transforms.v2.functional import cvcuda_to_tensor, to_cvcuda_tensor, to_image, to_pil_image
+from torchvision.transforms.v2.functional._utils import _import_cvcuda, _is_cvcuda_available
 from torchvision.utils import _Image_fromarray
 
 
 IN_OSS_CI = any(os.getenv(var) == "true" for var in ["CIRCLECI", "GITHUB_ACTIONS"])
 IN_RE_WORKER = os.environ.get("INSIDE_RE_WORKER") is not None
 IN_FBCODE = os.environ.get("IN_FBCODE_TORCHVISION") == "1"
+CVCUDA_AVAILABLE = _is_cvcuda_available()
 CUDA_NOT_AVAILABLE_MSG = "CUDA device not available"
 MPS_NOT_AVAILABLE_MSG = "MPS device not available"
 OSS_CI_GPU_NO_CUDA_MSG = "We're in an OSS GPU machine, and this test doesn't need cuda."
@@ -275,6 +277,17 @@ def combinations_grid(**kwargs):
     return [dict(zip(kwargs.keys(), values)) for values in itertools.product(*kwargs.values())]
 
 
+def cvcuda_to_pil_compatible_tensor(tensor):
+    tensor = cvcuda_to_tensor(tensor)
+    if tensor.ndim != 4:
+        raise ValueError(f"CV-CUDA Tensor should be 4 dimensional. Got {tensor.ndim} dimensions.")
+    if tensor.shape[0] != 1:
+        raise ValueError(
+            f"CV-CUDA Tensor should have batch dimension 1 for comparison with PIL.Image.Image. Got {tensor.shape[0]}."
+        )
+    return tensor.squeeze(0).cpu()
+
+
 class ImagePair(TensorLikePair):
     def __init__(
         self,
@@ -287,6 +300,16 @@ def __init__(
         if all(isinstance(input, PIL.Image.Image) for input in [actual, expected]):
             actual, expected = (to_image(input) for input in [actual, expected])
 
+        if CVCUDA_AVAILABLE and all(isinstance(input, _import_cvcuda().Tensor) for input in [actual, expected]):
+            actual, expected = (cvcuda_to_tensor(input) for input in [actual, expected])
+
+        if CVCUDA_AVAILABLE and isinstance(actual, _import_cvcuda().Tensor) and isinstance(expected, PIL.Image.Image):
+            actual = cvcuda_to_pil_compatible_tensor(actual)
+            expected = to_image(expected)
+
+        if CVCUDA_AVAILABLE and isinstance(actual, _import_cvcuda().Tensor):
+            actual = cvcuda_to_pil_compatible_tensor(actual)
+
         super().__init__(actual, expected, **other_parameters)
         self.mae = mae
 

test/test_transforms_v2.py

@@ -24,6 +24,7 @@
     assert_equal,
     cache,
     cpu_and_cuda,
+    cvcuda_to_pil_compatible_tensor,
     freeze_rng_state,
     ignore_jit_no_profile_information_warning,
     make_bounding_boxes,
@@ -6427,11 +6428,7 @@ def test_image_correctness(self, num_output_channels, color_space, make_input, f
         actual = fn(image, num_output_channels=num_output_channels)
 
         if make_input is make_image_cvcuda:
-            actual = F.cvcuda_to_tensor(actual).to(device="cpu")
-            actual = actual.squeeze(0)
-            # drop the batch dimension
-            image = F.cvcuda_to_tensor(image).to(device="cpu")
-            image = image.squeeze(0)
+            image = cvcuda_to_pil_compatible_tensor(image)
 
         expected = F.to_image(F.rgb_to_grayscale(F.to_pil_image(image), num_output_channels=num_output_channels))
 
@@ -6532,11 +6529,7 @@ def test_image_correctness(self, make_input, fn):
         actual = fn(image)
 
         if make_input is make_image_cvcuda:
-            actual = F.cvcuda_to_tensor(actual).to(device="cpu")
-            actual = actual.squeeze(0)
-            # drop the batch dimension
-            image = F.cvcuda_to_tensor(image).to(device="cpu")
-            image = image.squeeze(0)
+            image = cvcuda_to_pil_compatible_tensor(image)
 
         expected = F.to_image(F.grayscale_to_rgb(F.to_pil_image(image)))