Cleanup conversion transforms (#6801)

* remove copy from convert_color_space

* remove copy from convert_format_bounding_box

* remove .to_* methods from features

* remove unnecessary clones

* add perf todos

* refactor convert_color_space

* lint

* remove another clone

* and another clone

* remove a missed copy

Author: pmeier

test/prototype_transforms_kernel_infos.py

@@ -461,9 +461,7 @@ def transform(bbox):
             ],
             dtype=bbox.dtype,
         )
-        return F.convert_format_bounding_box(
-            out_bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
-        )
+        return F.convert_format_bounding_box(out_bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=format)
 
     if bounding_box.ndim < 2:
         bounding_box = [bounding_box]
@@ -556,26 +554,20 @@ def sample_inputs_affine_video():
 
 
 def sample_inputs_convert_format_bounding_box():
-    formats = set(features.BoundingBoxFormat)
-    for bounding_box_loader in make_bounding_box_loaders(formats=formats):
-        old_format = bounding_box_loader.format
-        for params in combinations_grid(new_format=formats - {old_format}, copy=(True, False)):
-            yield ArgsKwargs(bounding_box_loader, old_format=old_format, **params)
-
+    formats = list(features.BoundingBoxFormat)
+    for bounding_box_loader, new_format in itertools.product(make_bounding_box_loaders(formats=formats), formats):
+        yield ArgsKwargs(bounding_box_loader, old_format=bounding_box_loader.format, new_format=new_format)
 
-def reference_convert_format_bounding_box(bounding_box, old_format, new_format, copy):
-    if not copy:
-        raise pytest.UsageError("Reference for `convert_format_bounding_box` only supports `copy=True`")
 
+def reference_convert_format_bounding_box(bounding_box, old_format, new_format):
     return torchvision.ops.box_convert(
         bounding_box, in_fmt=old_format.kernel_name.lower(), out_fmt=new_format.kernel_name.lower()
     )
 
 
 def reference_inputs_convert_format_bounding_box():
     for args_kwargs in sample_inputs_convert_color_space_image_tensor():
-        (image_loader, *other_args), kwargs = args_kwargs
-        if len(image_loader.shape) == 2 and kwargs.setdefault("copy", True):
+        if len(args_kwargs.args[0].shape) == 2:
             yield args_kwargs
 
 
@@ -600,19 +592,19 @@ def sample_inputs_convert_color_space_image_tensor():
         for image_loader in make_image_loaders(
             sizes=["random"], color_spaces=[color_space], dtypes=[torch.float32], constant_alpha=True
         ):
-            yield ArgsKwargs(image_loader, old_color_space=color_space, new_color_space=color_space, copy=False)
+            yield ArgsKwargs(image_loader, old_color_space=color_space, new_color_space=color_space)
 
 
 @pil_reference_wrapper
-def reference_convert_color_space_image_tensor(image_pil, old_color_space, new_color_space, copy=True):
+def reference_convert_color_space_image_tensor(image_pil, old_color_space, new_color_space):
     color_space_pil = features.ColorSpace.from_pil_mode(image_pil.mode)
     if color_space_pil != old_color_space:
         raise pytest.UsageError(
             f"Converting the tensor image into an PIL image changed the colorspace "
             f"from {old_color_space} to {color_space_pil}"
         )
 
-    return F.convert_color_space_image_pil(image_pil, color_space=new_color_space, copy=copy)
+    return F.convert_color_space_image_pil(image_pil, color_space=new_color_space)
 
 
 def reference_inputs_convert_color_space_image_tensor():

test/test_prototype_transforms_functional.py

@@ -478,9 +478,7 @@ def _compute_expected_bbox(bbox, angle_, expand_, center_):
             device=bbox.device,
         )
         return (
-            convert_format_bounding_box(
-                out_bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox.format, copy=False
-            ),
+            convert_format_bounding_box(out_bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox.format),
             (height, width),
         )
 
@@ -733,14 +731,16 @@ def _compute_expected_bbox(bbox, padding_):
 
         bbox_format = bbox.format
         bbox_dtype = bbox.dtype
-        bbox = convert_format_bounding_box(bbox, old_format=bbox_format, new_format=features.BoundingBoxFormat.XYXY)
+        bbox = (
+            bbox.clone()
+            if bbox_format == features.BoundingBoxFormat.XYXY
+            else convert_format_bounding_box(bbox, bbox_format, features.BoundingBoxFormat.XYXY)
+        )
 
         bbox[0::2] += pad_left
         bbox[1::2] += pad_up
 
-        bbox = convert_format_bounding_box(
-            bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox_format, copy=False
-        )
+        bbox = convert_format_bounding_box(bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox_format)
         if bbox.dtype != bbox_dtype:
             # Temporary cast to original dtype
             # e.g. float32 -> int
@@ -840,9 +840,7 @@ def _compute_expected_bbox(bbox, pcoeffs_):
             dtype=bbox.dtype,
             device=bbox.device,
         )
-        return convert_format_bounding_box(
-            out_bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox.format, copy=False
-        )
+        return convert_format_bounding_box(out_bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox.format)
 
     spatial_size = (32, 38)
 
@@ -903,7 +901,7 @@ def _compute_expected_bbox(bbox, output_size_):
             dtype=bbox.dtype,
             device=bbox.device,
         )
-        return convert_format_bounding_box(out_bbox, features.BoundingBoxFormat.XYWH, format_, copy=False)
+        return convert_format_bounding_box(out_bbox, features.BoundingBoxFormat.XYWH, format_)
 
     for bboxes in make_bounding_boxes(extra_dims=((4,),)):
         bboxes = bboxes.to(device)

torchvision/prototype/features/_image.py

@@ -110,18 +110,6 @@ def spatial_size(self) -> Tuple[int, int]:
     def num_channels(self) -> int:
         return self.shape[-3]
 
-    def to_color_space(self, color_space: Union[str, ColorSpace], copy: bool = True) -> Image:
-        if isinstance(color_space, str):
-            color_space = ColorSpace.from_str(color_space.upper())
-
-        return Image.wrap_like(
-            self,
-            self._F.convert_color_space_image_tensor(
-                self.as_subclass(torch.Tensor), old_color_space=self.color_space, new_color_space=color_space, copy=copy
-            ),
-            color_space=color_space,
-        )
-
     def horizontal_flip(self) -> Image:
         output = self._F.horizontal_flip_image_tensor(self.as_subclass(torch.Tensor))
         return Image.wrap_like(self, output)

torchvision/prototype/features/_video.py

@@ -66,18 +66,6 @@ def num_channels(self) -> int:
     def num_frames(self) -> int:
         return self.shape[-4]
 
-    def to_color_space(self, color_space: Union[str, ColorSpace], copy: bool = True) -> Video:
-        if isinstance(color_space, str):
-            color_space = ColorSpace.from_str(color_space.upper())
-
-        return Video.wrap_like(
-            self,
-            self._F.convert_color_space_video(
-                self.as_subclass(torch.Tensor), old_color_space=self.color_space, new_color_space=color_space, copy=copy
-            ),
-            color_space=color_space,
-        )
-
     def horizontal_flip(self) -> Video:
         output = self._F.horizontal_flip_video(self.as_subclass(torch.Tensor))
         return Video.wrap_like(self, output)

torchvision/prototype/transforms/_augment.py

@@ -265,7 +265,7 @@ def _copy_paste(
         # https://github.com/pytorch/vision/blob/b6feccbc4387766b76a3e22b13815dbbbfa87c0f/torchvision/models/detection/roi_heads.py#L418-L422
         xyxy_boxes[:, 2:] += 1
         boxes = F.convert_format_bounding_box(
-            xyxy_boxes, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox_format, copy=False
+            xyxy_boxes, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox_format
         )
         out_target["boxes"] = torch.cat([boxes, paste_boxes])
 

torchvision/prototype/transforms/_geometry.py

@@ -655,9 +655,7 @@ def _get_params(self, flat_inputs: List[Any]) -> Dict[str, Any]:
                     continue
 
                 # check for any valid boxes with centers within the crop area
-                xyxy_bboxes = F.convert_format_bounding_box(
-                    bboxes, old_format=bboxes.format, new_format=features.BoundingBoxFormat.XYXY, copy=True
-                )
+                xyxy_bboxes = F.convert_format_bounding_box(bboxes, bboxes.format, features.BoundingBoxFormat.XYXY)
                 cx = 0.5 * (xyxy_bboxes[..., 0] + xyxy_bboxes[..., 2])
                 cy = 0.5 * (xyxy_bboxes[..., 1] + xyxy_bboxes[..., 3])
                 is_within_crop_area = (left < cx) & (cx < right) & (top < cy) & (cy < bottom)
@@ -801,22 +799,21 @@ def _get_params(self, flat_inputs: List[Any]) -> Dict[str, Any]:
         top = int(offset_height * r)
         left = int(offset_width * r)
 
+        bounding_boxes: Optional[torch.Tensor]
         try:
             bounding_boxes = query_bounding_box(flat_inputs)
         except ValueError:
             bounding_boxes = None
 
         if needs_crop and bounding_boxes is not None:
-            bounding_boxes = cast(
-                features.BoundingBox, F.crop(bounding_boxes, top=top, left=left, height=new_height, width=new_width)
-            )
-            bounding_boxes = features.BoundingBox.wrap_like(
-                bounding_boxes,
-                F.clamp_bounding_box(
-                    bounding_boxes, format=bounding_boxes.format, spatial_size=bounding_boxes.spatial_size
-                ),
+            format = bounding_boxes.format
+            bounding_boxes, spatial_size = F.crop_bounding_box(
+                bounding_boxes, format=format, top=top, left=left, height=new_height, width=new_width
             )
-            height_and_width = bounding_boxes.to_format(features.BoundingBoxFormat.XYWH)[..., 2:]
+            bounding_boxes = F.clamp_bounding_box(bounding_boxes, format=format, spatial_size=spatial_size)
+            height_and_width = F.convert_format_bounding_box(
+                bounding_boxes, old_format=format, new_format=features.BoundingBoxFormat.XYWH
+            )[..., 2:]
             is_valid = torch.all(height_and_width > 0, dim=-1)
         else:
             is_valid = None

torchvision/prototype/transforms/_meta.py

@@ -50,7 +50,6 @@ def __init__(
         self,
         color_space: Union[str, features.ColorSpace],
         old_color_space: Optional[Union[str, features.ColorSpace]] = None,
-        copy: bool = True,
     ) -> None:
         super().__init__()
 
@@ -62,14 +61,10 @@ def __init__(
             old_color_space = features.ColorSpace.from_str(old_color_space)
         self.old_color_space = old_color_space
 
-        self.copy = copy
-
     def _transform(
         self, inpt: Union[features.ImageType, features.VideoType], params: Dict[str, Any]
     ) -> Union[features.ImageType, features.VideoType]:
-        return F.convert_color_space(
-            inpt, color_space=self.color_space, old_color_space=self.old_color_space, copy=self.copy
-        )
+        return F.convert_color_space(inpt, color_space=self.color_space, old_color_space=self.old_color_space)
 
 
 class ClampBoundingBoxes(Transform):

torchvision/prototype/transforms/functional/_geometry.py

@@ -36,14 +36,18 @@ def horizontal_flip_bounding_box(
 ) -> torch.Tensor:
     shape = bounding_box.shape
 
-    bounding_box = convert_format_bounding_box(
-        bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY
+    # TODO: Investigate if it makes sense from a performance perspective to have an implementation for every
+    #  BoundingBoxFormat instead of converting back and forth
+    bounding_box = (
+        bounding_box.clone()
+        if format == features.BoundingBoxFormat.XYXY
+        else convert_format_bounding_box(bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY)
     ).reshape(-1, 4)
 
     bounding_box[:, [0, 2]] = spatial_size[1] - bounding_box[:, [2, 0]]
 
     return convert_format_bounding_box(
-        bounding_box, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
+        bounding_box, old_format=features.BoundingBoxFormat.XYXY, new_format=format
     ).reshape(shape)
 
 
@@ -73,14 +77,18 @@ def vertical_flip_bounding_box(
 ) -> torch.Tensor:
     shape = bounding_box.shape
 
-    bounding_box = convert_format_bounding_box(
-        bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY
+    # TODO: Investigate if it makes sense from a performance perspective to have an implementation for every
+    #  BoundingBoxFormat instead of converting back and forth
+    bounding_box = (
+        bounding_box.clone()
+        if format == features.BoundingBoxFormat.XYXY
+        else convert_format_bounding_box(bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY)
     ).reshape(-1, 4)
 
     bounding_box[:, [1, 3]] = spatial_size[0] - bounding_box[:, [3, 1]]
 
     return convert_format_bounding_box(
-        bounding_box, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
+        bounding_box, old_format=features.BoundingBoxFormat.XYXY, new_format=format
     ).reshape(shape)
 
 
@@ -394,16 +402,17 @@ def affine_bounding_box(
     center: Optional[List[float]] = None,
 ) -> torch.Tensor:
     original_shape = bounding_box.shape
-    bounding_box = convert_format_bounding_box(
-        bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY
+
+    bounding_box = (
+        convert_format_bounding_box(bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY)
     ).reshape(-1, 4)
 
     out_bboxes, _ = _affine_bounding_box_xyxy(bounding_box, spatial_size, angle, translate, scale, shear, center)
 
     # out_bboxes should be of shape [N boxes, 4]
 
     return convert_format_bounding_box(
-        out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
+        out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format
     ).reshape(original_shape)
 
 
@@ -583,8 +592,8 @@ def rotate_bounding_box(
         center = None
 
     original_shape = bounding_box.shape
-    bounding_box = convert_format_bounding_box(
-        bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY
+    bounding_box = (
+        convert_format_bounding_box(bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY)
     ).reshape(-1, 4)
 
     out_bboxes, spatial_size = _affine_bounding_box_xyxy(
@@ -599,9 +608,9 @@ def rotate_bounding_box(
     )
 
     return (
-        convert_format_bounding_box(
-            out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
-        ).reshape(original_shape),
+        convert_format_bounding_box(out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format).reshape(
+            original_shape
+        ),
         spatial_size,
     )
 
@@ -818,18 +827,20 @@ def crop_bounding_box(
     height: int,
     width: int,
 ) -> Tuple[torch.Tensor, Tuple[int, int]]:
-    bounding_box = convert_format_bounding_box(
-        bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY
+    # TODO: Investigate if it makes sense from a performance perspective to have an implementation for every
+    #  BoundingBoxFormat instead of converting back and forth
+    bounding_box = (
+        bounding_box.clone()
+        if format == features.BoundingBoxFormat.XYXY
+        else convert_format_bounding_box(bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY)
     )
 
     # Crop or implicit pad if left and/or top have negative values:
     bounding_box[..., 0::2] -= left
     bounding_box[..., 1::2] -= top
 
     return (
-        convert_format_bounding_box(
-            bounding_box, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
-        ),
+        convert_format_bounding_box(bounding_box, old_format=features.BoundingBoxFormat.XYXY, new_format=format),
         (height, width),
     )
 
@@ -896,8 +907,8 @@ def perspective_bounding_box(
         raise ValueError("Argument perspective_coeffs should have 8 float values")
 
     original_shape = bounding_box.shape
-    bounding_box = convert_format_bounding_box(
-        bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY
+    bounding_box = (
+        convert_format_bounding_box(bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY)
     ).reshape(-1, 4)
 
     dtype = bounding_box.dtype if torch.is_floating_point(bounding_box) else torch.float32
@@ -967,7 +978,7 @@ def perspective_bounding_box(
     # out_bboxes should be of shape [N boxes, 4]
 
     return convert_format_bounding_box(
-        out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
+        out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format
     ).reshape(original_shape)
 
 
@@ -1061,8 +1072,8 @@ def elastic_bounding_box(
     displacement = displacement.to(bounding_box.device)
 
     original_shape = bounding_box.shape
-    bounding_box = convert_format_bounding_box(
-        bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY
+    bounding_box = (
+        convert_format_bounding_box(bounding_box, old_format=format, new_format=features.BoundingBoxFormat.XYXY)
     ).reshape(-1, 4)
 
     # Question (vfdev-5): should we rely on good displacement shape and fetch image size from it
@@ -1088,7 +1099,7 @@ def elastic_bounding_box(
     out_bboxes = torch.cat([out_bbox_mins, out_bbox_maxs], dim=1).to(bounding_box.dtype)
 
     return convert_format_bounding_box(
-        out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format, copy=False
+        out_bboxes, old_format=features.BoundingBoxFormat.XYXY, new_format=format
     ).reshape(original_shape)