[fbsync] Added center arg to F.affine and RandomAffine ops (#5208)

Summary:
* Added center option to F.affine and RandomAffine ops

* Updates according to the review

Reviewed By: kazhang

Differential Revision: D33927487

fbshipit-source-id: 3e6562977c7948331982fa5f788f7933744917ac

Author: yiwen-song

test/test_functional_tensor.py

@@ -232,7 +232,8 @@ def test_square_rotations(self, device, height, width, dt, angle, config, fn):
     @pytest.mark.parametrize("dt", ALL_DTYPES)
     @pytest.mark.parametrize("angle", [90, 45, 15, -30, -60, -120])
     @pytest.mark.parametrize("fn", [F.affine, scripted_affine])
-    def test_rect_rotations(self, device, height, width, dt, angle, fn):
+    @pytest.mark.parametrize("center", [None, [0, 0]])
+    def test_rect_rotations(self, device, height, width, dt, angle, fn, center):
         # Tests on rectangular images
         tensor, pil_img = _create_data(height, width, device=device)
 
@@ -244,11 +245,13 @@ def test_rect_rotations(self, device, height, width, dt, angle, fn):
             tensor = tensor.to(dtype=dt)
 
         out_pil_img = F.affine(
-            pil_img, angle=angle, translate=[0, 0], scale=1.0, shear=[0.0, 0.0], interpolation=NEAREST
+            pil_img, angle=angle, translate=[0, 0], scale=1.0, shear=[0.0, 0.0], interpolation=NEAREST, center=center
         )
         out_pil_tensor = torch.from_numpy(np.array(out_pil_img).transpose((2, 0, 1)))
 
-        out_tensor = fn(tensor, angle=angle, translate=[0, 0], scale=1.0, shear=[0.0, 0.0], interpolation=NEAREST).cpu()
+        out_tensor = fn(
+            tensor, angle=angle, translate=[0, 0], scale=1.0, shear=[0.0, 0.0], interpolation=NEAREST, center=center
+        ).cpu()
 
         if out_tensor.dtype != torch.uint8:
             out_tensor = out_tensor.to(torch.uint8)

test/test_transforms.py

@@ -1983,11 +1983,11 @@ def _to_3x3_inv(self, inv_result_matrix):
         result_matrix[2, 2] = 1
         return np.linalg.inv(result_matrix)
 
-    def _test_transformation(self, angle, translate, scale, shear, pil_image, input_img):
+    def _test_transformation(self, angle, translate, scale, shear, pil_image, input_img, center=None):
 
         a_rad = math.radians(angle)
         s_rad = [math.radians(sh_) for sh_ in shear]
-        cnt = [20, 20]
+        cnt = [20, 20] if center is None else center
         cx, cy = cnt
         tx, ty = translate
         sx, sy = s_rad
@@ -2032,7 +2032,7 @@ def _test_transformation(self, angle, translate, scale, shear, pil_image, input_
                 if 0 <= _x < input_img.shape[1] and 0 <= _y < input_img.shape[0]:
                     true_result[y, x, :] = input_img[_y, _x, :]
 
-        result = F.affine(pil_image, angle=angle, translate=translate, scale=scale, shear=shear)
+        result = F.affine(pil_image, angle=angle, translate=translate, scale=scale, shear=shear, center=center)
         assert result.size == pil_image.size
         # Compute number of different pixels:
         np_result = np.array(result)
@@ -2050,6 +2050,18 @@ def test_transformation_discrete(self, pil_image, input_img):
             angle=angle, translate=(0, 0), scale=1.0, shear=(0.0, 0.0), pil_image=pil_image, input_img=input_img
         )
 
+        # Test rotation
+        angle = 45
+        self._test_transformation(
+            angle=angle,
+            translate=(0, 0),
+            scale=1.0,
+            shear=(0.0, 0.0),
+            pil_image=pil_image,
+            input_img=input_img,
+            center=[0, 0],
+        )
+
         # Test translation
         translate = [10, 15]
         self._test_transformation(
@@ -2068,6 +2080,18 @@ def test_transformation_discrete(self, pil_image, input_img):
             angle=0.0, translate=(0.0, 0.0), scale=1.0, shear=shear, pil_image=pil_image, input_img=input_img
         )
 
+        # Test shear with top-left as center
+        shear = [45.0, 25.0]
+        self._test_transformation(
+            angle=0.0,
+            translate=(0.0, 0.0),
+            scale=1.0,
+            shear=shear,
+            pil_image=pil_image,
+            input_img=input_img,
+            center=[0, 0],
+        )
+
     @pytest.mark.parametrize("angle", range(-90, 90, 36))
     @pytest.mark.parametrize("translate", range(-10, 10, 5))
     @pytest.mark.parametrize("scale", [0.77, 1.0, 1.27])

torchvision/transforms/functional.py

@@ -945,26 +945,31 @@ def adjust_gamma(img: Tensor, gamma: float, gain: float = 1) -> Tensor:
 
 
 def _get_inverse_affine_matrix(
-    center: List[float], angle: float, translate: List[float], scale: float, shear: List[float]
+    center: List[float],
+    angle: float,
+    translate: List[float],
+    scale: float,
+    shear: List[float],
 ) -> List[float]:
     # Helper method to compute inverse matrix for affine transformation
 
-    # As it is explained in PIL.Image.rotate
-    # We need compute INVERSE of affine transformation matrix: M = T * C * RSS * C^-1
+    # Pillow requires inverse affine transformation matrix:
+    # Affine matrix is : M = T * C * RotateScaleShear * C^-1
+    #
     # where T is translation matrix: [1, 0, tx | 0, 1, ty | 0, 0, 1]
     #       C is translation matrix to keep center: [1, 0, cx | 0, 1, cy | 0, 0, 1]
-    #       RSS is rotation with scale and shear matrix
-    #       RSS(a, s, (sx, sy)) =
+    #       RotateScaleShear is rotation with scale and shear matrix
+    #
+    #       RotateScaleShear(a, s, (sx, sy)) =
     #       = R(a) * S(s) * SHy(sy) * SHx(sx)
-    #       = [ s*cos(a - sy)/cos(sy), s*(-cos(a - sy)*tan(x)/cos(y) - sin(a)), 0 ]
-    #         [ s*sin(a + sy)/cos(sy), s*(-sin(a - sy)*tan(x)/cos(y) + cos(a)), 0 ]
+    #       = [ s*cos(a - sy)/cos(sy), s*(-cos(a - sy)*tan(sx)/cos(sy) - sin(a)), 0 ]
+    #         [ s*sin(a + sy)/cos(sy), s*(-sin(a - sy)*tan(sx)/cos(sy) + cos(a)), 0 ]
     #         [ 0                    , 0                                      , 1 ]
-    #
     # where R is a rotation matrix, S is a scaling matrix, and SHx and SHy are the shears:
     # SHx(s) = [1, -tan(s)] and SHy(s) = [1      , 0]
     #          [0, 1      ]              [-tan(s), 1]
     #
-    # Thus, the inverse is M^-1 = C * RSS^-1 * C^-1 * T^-1
+    # Thus, the inverse is M^-1 = C * RotateScaleShear^-1 * C^-1 * T^-1
 
     rot = math.radians(angle)
     sx = math.radians(shear[0])
@@ -1085,6 +1090,7 @@ def affine(
     fill: Optional[List[float]] = None,
     resample: Optional[int] = None,
     fillcolor: Optional[List[float]] = None,
+    center: Optional[List[int]] = None,
 ) -> Tensor:
     """Apply affine transformation on the image keeping image center invariant.
     If the image is torch Tensor, it is expected
@@ -1112,6 +1118,8 @@ def affine(
             Please use the ``fill`` parameter instead.
         resample (int, optional): deprecated argument and will be removed since v0.10.0.
             Please use the ``interpolation`` parameter instead.
+        center (sequence, optional): Optional center of rotation. Origin is the upper left corner.
+            Default is the center of the image.
 
     Returns:
         PIL Image or Tensor: Transformed image.
@@ -1172,18 +1180,28 @@ def affine(
     if len(shear) != 2:
         raise ValueError(f"Shear should be a sequence containing two values. Got {shear}")
 
+    if center is not None and not isinstance(center, (list, tuple)):
+        raise TypeError("Argument center should be a sequence")
+
     img_size = get_image_size(img)
     if not isinstance(img, torch.Tensor):
         # center = (img_size[0] * 0.5 + 0.5, img_size[1] * 0.5 + 0.5)
         # it is visually better to estimate the center without 0.5 offset
         # otherwise image rotated by 90 degrees is shifted vs output image of torch.rot90 or F_t.affine
-        center = [img_size[0] * 0.5, img_size[1] * 0.5]
+        if center is None:
+            center = [img_size[0] * 0.5, img_size[1] * 0.5]
         matrix = _get_inverse_affine_matrix(center, angle, translate, scale, shear)
         pil_interpolation = pil_modes_mapping[interpolation]
         return F_pil.affine(img, matrix=matrix, interpolation=pil_interpolation, fill=fill)
 
+    center_f = [0.0, 0.0]
+    if center is not None:
+        img_size = get_image_size(img)
+        # Center values should be in pixel coordinates but translated such that (0, 0) corresponds to image center.
+        center_f = [1.0 * (c - s * 0.5) for c, s in zip(center, img_size)]
+
     translate_f = [1.0 * t for t in translate]
-    matrix = _get_inverse_affine_matrix([0.0, 0.0], angle, translate_f, scale, shear)
+    matrix = _get_inverse_affine_matrix(center_f, angle, translate_f, scale, shear)
     return F_t.affine(img, matrix=matrix, interpolation=interpolation.value, fill=fill)
 
 

torchvision/transforms/transforms.py

@@ -1414,6 +1414,8 @@ class RandomAffine(torch.nn.Module):
             Please use the ``fill`` parameter instead.
         resample (int, optional): deprecated argument and will be removed since v0.10.0.
             Please use the ``interpolation`` parameter instead.
+        center (sequence, optional): Optional center of rotation, (x, y). Origin is the upper left corner.
+            Default is the center of the image.
 
     .. _filters: https://pillow.readthedocs.io/en/latest/handbook/concepts.html#filters
 
@@ -1429,6 +1431,7 @@ def __init__(
         fill=0,
         fillcolor=None,
         resample=None,
+        center=None,
     ):
         super().__init__()
         _log_api_usage_once(self)
@@ -1482,6 +1485,11 @@ def __init__(
 
         self.fillcolor = self.fill = fill
 
+        if center is not None:
+            _check_sequence_input(center, "center", req_sizes=(2,))
+
+        self.center = center
+
     @staticmethod
     def get_params(
         degrees: List[float],
@@ -1538,7 +1546,7 @@ def forward(self, img):
 
         ret = self.get_params(self.degrees, self.translate, self.scale, self.shear, img_size)
 
-        return F.affine(img, *ret, interpolation=self.interpolation, fill=fill)
+        return F.affine(img, *ret, interpolation=self.interpolation, fill=fill, center=self.center)
 
     def __repr__(self):
         s = "{name}(degrees={degrees}"
@@ -1552,6 +1560,8 @@ def __repr__(self):
             s += ", interpolation={interpolation}"
         if self.fill != 0:
             s += ", fill={fill}"
+        if self.center is not None:
+            s += ", center={center}"
         s += ")"
         d = dict(self.__dict__)
         d["interpolation"] = self.interpolation.value