support grayscale / RGB alpha conversions (#5567)

* support grayscale / RGB alpha conversions

* use _max_valu from stable

* remove extra copy for PIL conversion

* simplify test image generation for color spaces with alpha channel

* use common _max_value in tests

* replace dynamically created dicts with if/else

* make color space conversion more explicit

* make even more explicit

* simplify alpha image generation

* fix if / elif

* add error for unknown conversions

* rename RGBA to RGB_ALPHA

* cleanup

* GRAYSCALE to GRAY

test/test_prototype_transforms.py

@@ -102,7 +102,14 @@ class TestSmoke:
             (
                 transform,
                 itertools.chain.from_iterable(
-                    fn(dtypes=[torch.uint8], extra_dims=[(4,)])
+                    fn(
+                        color_spaces=[
+                            features.ColorSpace.GRAY,
+                            features.ColorSpace.RGB,
+                        ],
+                        dtypes=[torch.uint8],
+                        extra_dims=[(4,)],
+                    )
                     for fn in [
                         make_images,
                         make_vanilla_tensor_images,
@@ -152,3 +159,32 @@ class TestSmoke:
     )
     def test_random_resized_crop(self, transform, input):
         transform(input)
+
+    @parametrize(
+        [
+            (
+                transforms.ConvertImageColorSpace(color_space=new_color_space, old_color_space=old_color_space),
+                itertools.chain.from_iterable(
+                    [
+                        fn(color_spaces=[old_color_space])
+                        for fn in (
+                            make_images,
+                            make_vanilla_tensor_images,
+                            make_pil_images,
+                        )
+                    ]
+                ),
+            )
+            for old_color_space, new_color_space in itertools.product(
+                [
+                    features.ColorSpace.GRAY,
+                    features.ColorSpace.GRAY_ALPHA,
+                    features.ColorSpace.RGB,
+                    features.ColorSpace.RGB_ALPHA,
+                ],
+                repeat=2,
+            )
+        ]
+    )
+    def test_convert_image_color_space(self, transform, input):
+        transform(input)

test/test_prototype_transforms_functional.py

@@ -7,33 +7,44 @@ import torchvision.prototype.transforms.functional as F
 from torch import jit
 from torch.nn.functional import one_hot
 from torchvision.prototype import features
+from torchvision.transforms.functional_tensor import _max_value as get_max_value
 
 make_tensor = functools.partial(torch.testing.make_tensor, device="cpu")
 
 
-def make_image(size=None, *, color_space, extra_dims=(), dtype=torch.float32):
+def make_image(size=None, *, color_space, extra_dims=(), dtype=torch.float32, constant_alpha=True):
     size = size or torch.randint(16, 33, (2,)).tolist()
 
-    num_channels = {
-        features.ColorSpace.GRAYSCALE: 1,
-        features.ColorSpace.RGB: 3,
-    }[color_space]
+    try:
+        num_channels = {
+            features.ColorSpace.GRAY: 1,
+            features.ColorSpace.GRAY_ALPHA: 2,
+            features.ColorSpace.RGB: 3,
+            features.ColorSpace.RGB_ALPHA: 4,
+        }[color_space]
+    except KeyError as error:
+        raise pytest.UsageError() from error
 
     shape = (*extra_dims, num_channels, *size)
-    if dtype.is_floating_point:
-        data = torch.rand(shape, dtype=dtype)
-    else:
-        data = torch.randint(0, torch.iinfo(dtype).max, shape, dtype=dtype)
+    max_value = get_max_value(dtype)
+    data = make_tensor(shape, low=0, high=max_value, dtype=dtype)
+    if color_space in {features.ColorSpace.GRAY_ALPHA, features.ColorSpace.RGB_ALPHA} and constant_alpha:
+        data[..., -1, :, :] = max_value
     return features.Image(data, color_space=color_space)
 
 
-make_grayscale_image = functools.partial(make_image, color_space=features.ColorSpace.GRAYSCALE)
+make_grayscale_image = functools.partial(make_image, color_space=features.ColorSpace.GRAY)
 make_rgb_image = functools.partial(make_image, color_space=features.ColorSpace.RGB)
 
 
 def make_images(
     sizes=((16, 16), (7, 33), (31, 9)),
-    color_spaces=(features.ColorSpace.GRAYSCALE, features.ColorSpace.RGB),
+    color_spaces=(
+        features.ColorSpace.GRAY,
+        features.ColorSpace.GRAY_ALPHA,
+        features.ColorSpace.RGB,
+        features.ColorSpace.RGB_ALPHA,
+    ),
     dtypes=(torch.float32, torch.uint8),
     extra_dims=((4,), (2, 3)),
 ):
@@ -48,15 +59,12 @@ def randint_with_tensor_bounds(arg1, arg2=None, **kwargs):
     low, high = torch.broadcast_tensors(
         *[torch.as_tensor(arg) for arg in ((0, arg1) if arg2 is None else (arg1, arg2))]
     )
-    try:
-        return torch.stack(
-            [
-                torch.randint(low_scalar, high_scalar, (), **kwargs)
-                for low_scalar, high_scalar in zip(low.flatten().tolist(), high.flatten().tolist())
-            ]
-        ).reshape(low.shape)
-    except RuntimeError as error:
-        raise error
+    return torch.stack(
+        [
+            torch.randint(low_scalar, high_scalar, (), **kwargs)
+            for low_scalar, high_scalar in zip(low.flatten().tolist(), high.flatten().tolist())
+        ]
+    ).reshape(low.shape)
 
 
 def make_bounding_box(*, format, image_size=(32, 32), extra_dims=(), dtype=torch.int64):
@@ -83,8 +91,8 @@ def make_bounding_box(*, format, image_size=(32, 32), extra_dims=(), dtype=torch
         w = randint_with_tensor_bounds(1, torch.minimum(cx, width - cx) + 1)
         h = randint_with_tensor_bounds(1, torch.minimum(cy, width - cy) + 1)
         parts = (cx, cy, w, h)
-    else:  # format == features.BoundingBoxFormat._SENTINEL:
-        raise ValueError()
+    else:
+        raise pytest.UsageError()
 
     return features.BoundingBox(torch.stack(parts, dim=-1).to(dtype), format=format, image_size=image_size)
 

torchvision/prototype/features/_image.py

@@ -15,8 +15,23 @@ from ._feature import _Feature
 
 class ColorSpace(StrEnum):
     OTHER = StrEnum.auto()
-    GRAYSCALE = StrEnum.auto()
+    GRAY = StrEnum.auto()
+    GRAY_ALPHA = StrEnum.auto()
     RGB = StrEnum.auto()
+    RGB_ALPHA = StrEnum.auto()
+
+    @classmethod
+    def from_pil_mode(cls, mode: str) -> ColorSpace:
+        if mode == "L":
+            return cls.GRAY
+        elif mode == "LA":
+            return cls.GRAY_ALPHA
+        elif mode == "RGB":
+            return cls.RGB
+        elif mode == "RGBA":
+            return cls.RGB_ALPHA
+        else:
+            return cls.OTHER
 
 
 class Image(_Feature):
@@ -71,13 +86,17 @@ class Image(_Feature):
         if data.ndim < 2:
             return ColorSpace.OTHER
         elif data.ndim == 2:
-            return ColorSpace.GRAYSCALE
+            return ColorSpace.GRAY
 
         num_channels = data.shape[-3]
         if num_channels == 1:
-            return ColorSpace.GRAYSCALE
+            return ColorSpace.GRAY
+        elif num_channels == 2:
+            return ColorSpace.GRAY_ALPHA
         elif num_channels == 3:
             return ColorSpace.RGB
+        elif num_channels == 4:
+            return ColorSpace.RGB_ALPHA
         else:
             return ColorSpace.OTHER
 

torchvision/prototype/transforms/_meta.py

@@ -48,11 +48,11 @@ class ConvertImageColorSpace(Transform):
         super().__init__()
 
         if isinstance(color_space, str):
-            color_space = features.ColorSpace[color_space]
+            color_space = features.ColorSpace.from_str(color_space)
         self.color_space = color_space
 
         if isinstance(old_color_space, str):
-            old_color_space = features.ColorSpace[old_color_space]
+            old_color_space = features.ColorSpace.from_str(old_color_space)
         self.old_color_space = old_color_space
 
     def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
@@ -72,13 +72,6 @@ class ConvertImageColorSpace(Transform):
                 input, old_color_space=self.old_color_space, new_color_space=self.color_space
             )
         elif isinstance(input, PIL.Image.Image):
-            old_color_space = {
-                "L": features.ColorSpace.GRAYSCALE,
-                "RGB": features.ColorSpace.RGB,
-            }.get(input.mode, features.ColorSpace.OTHER)
-
-            return F.convert_image_color_space_pil(
-                input, old_color_space=old_color_space, new_color_space=self.color_space
-            )
+            return F.convert_image_color_space_pil(input, color_space=self.color_space)
         else:
             return input

torchvision/prototype/transforms/functional/_meta.py

+from typing import Tuple, Optional
+
 import PIL.Image
 import torch
 from torchvision.prototype.features import BoundingBoxFormat, ColorSpace
 from torchvision.transforms import functional_tensor as _FT, functional_pil as _FP
 
-
 get_dimensions_image_tensor = _FT.get_dimensions
 get_dimensions_image_pil = _FP.get_dimensions
 
@@ -57,41 +58,88 @@ def convert_bounding_box_format(
     return bounding_box
 
 
-def _grayscale_to_rgb_tensor(grayscale: torch.Tensor) -> torch.Tensor:
-    repeats = [1] * grayscale.ndim
-    repeats[-3] = 3
-    return grayscale.repeat(repeats)
+def _split_alpha(image: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+    return image[..., :-1, :, :], image[..., -1:, :, :]
 
 
-def convert_image_color_space_tensor(
-    image: torch.Tensor, old_color_space: ColorSpace, new_color_space: ColorSpace
-) -> torch.Tensor:
-    if new_color_space == old_color_space:
-        return image.clone()
+def _strip_alpha(image: torch.Tensor) -> torch.Tensor:
+    image, alpha = _split_alpha(image)
+    if not torch.all(alpha == _FT._max_value(alpha.dtype)):
+        raise RuntimeError(
+            "Stripping the alpha channel if it contains values other than the max value is not supported."
+        )
+    return image
 
-    if old_color_space == ColorSpace.GRAYSCALE:
-        image = _grayscale_to_rgb_tensor(image)
 
-    if new_color_space == ColorSpace.GRAYSCALE:
-        image = _FT.rgb_to_grayscale(image)
+def _add_alpha(image: torch.Tensor, alpha: Optional[torch.Tensor] = None) -> torch.Tensor:
+    if alpha is None:
+        shape = list(image.shape)
+        shape[-3] = 1
+        alpha = torch.full(shape, _FT._max_value(image.dtype), dtype=image.dtype, device=image.device)
+    return torch.cat((image, alpha), dim=-3)
 
-    return image
 
-
-def _grayscale_to_rgb_pil(grayscale: PIL.Image.Image) -> PIL.Image.Image:
-    return grayscale.convert("RGB")
+def _gray_to_rgb(grayscale: torch.Tensor) -> torch.Tensor:
+    repeats = [1] * grayscale.ndim
+    repeats[-3] = 3
+    return grayscale.repeat(repeats)
 
 
-def convert_image_color_space_pil(
-    image: PIL.Image.Image, old_color_space: ColorSpace, new_color_space: ColorSpace
-) -> PIL.Image.Image:
-    if new_color_space == old_color_space:
-        return image.copy()
+_rgb_to_gray = _FT.rgb_to_grayscale
 
-    if old_color_space == ColorSpace.GRAYSCALE:
-        image = _grayscale_to_rgb_pil(image)
 
-    if new_color_space == ColorSpace.GRAYSCALE:
-        image = _FP.to_grayscale(image)
+def convert_image_color_space_tensor(
+    image: torch.Tensor, old_color_space: ColorSpace, new_color_space: ColorSpace
+) -> torch.Tensor:
+    if new_color_space == old_color_space:
+        return image.clone()
 
-    return image
+    if old_color_space == ColorSpace.OTHER or new_color_space == ColorSpace.OTHER:
+        raise RuntimeError(f"Conversion to or from {ColorSpace.OTHER} is not supported.")
+
+    if old_color_space == ColorSpace.GRAY and new_color_space == ColorSpace.GRAY_ALPHA:
+        return _add_alpha(image)
+    elif old_color_space == ColorSpace.GRAY and new_color_space == ColorSpace.RGB:
+        return _gray_to_rgb(image)
+    elif old_color_space == ColorSpace.GRAY and new_color_space == ColorSpace.RGB_ALPHA:
+        return _add_alpha(_gray_to_rgb(image))
+    elif old_color_space == ColorSpace.GRAY_ALPHA and new_color_space == ColorSpace.GRAY:
+        return _strip_alpha(image)
+    elif old_color_space == ColorSpace.GRAY_ALPHA and new_color_space == ColorSpace.RGB:
+        return _gray_to_rgb(_strip_alpha(image))
+    elif old_color_space == ColorSpace.GRAY_ALPHA and new_color_space == ColorSpace.RGB_ALPHA:
+        image, alpha = _split_alpha(image)
+        return _add_alpha(_gray_to_rgb(image), alpha)
+    elif old_color_space == ColorSpace.RGB and new_color_space == ColorSpace.GRAY:
+        return _rgb_to_gray(image)
+    elif old_color_space == ColorSpace.RGB and new_color_space == ColorSpace.GRAY_ALPHA:
+        return _add_alpha(_rgb_to_gray(image))
+    elif old_color_space == ColorSpace.RGB and new_color_space == ColorSpace.RGB_ALPHA:
+        return _add_alpha(image)
+    elif old_color_space == ColorSpace.RGB_ALPHA and new_color_space == ColorSpace.GRAY:
+        return _rgb_to_gray(_strip_alpha(image))
+    elif old_color_space == ColorSpace.RGB_ALPHA and new_color_space == ColorSpace.GRAY_ALPHA:
+        image, alpha = _split_alpha(image)
+        return _add_alpha(_rgb_to_gray(image), alpha)
+    elif old_color_space == ColorSpace.RGB_ALPHA and new_color_space == ColorSpace.RGB:
+        return _strip_alpha(image)
+    else:
+        raise RuntimeError(f"Conversion from {old_color_space} to {new_color_space} is not supported.")
+
+
+_COLOR_SPACE_TO_PIL_MODE = {
+    ColorSpace.GRAY: "L",
+    ColorSpace.GRAY_ALPHA: "LA",
+    ColorSpace.RGB: "RGB",
+    ColorSpace.RGB_ALPHA: "RGBA",
+}
+
+
+def convert_image_color_space_pil(image: PIL.Image.Image, color_space: ColorSpace) -> PIL.Image.Image:
+    old_mode = image.mode
+    try:
+        new_mode = _COLOR_SPACE_TO_PIL_MODE[color_space]
+    except KeyError:
+        raise ValueError(f"Conversion from {ColorSpace.from_pil_mode(old_mode)} to {color_space} is not supported.")
+
+    return image.convert(new_mode)