Improved docs and tests for (#2371)

- RandomCrop: crop with padding using all commonly supported modes

test/test_transforms_tensor.py

@@ -26,24 +26,29 @@ class Tester(unittest.TestCase):
         transformed_pil_img = getattr(F, func)(pil_img, **fn_kwargs)
         self.compareTensorToPIL(transformed_tensor, transformed_pil_img)
 
-    def _test_geom_op(self, func, method, fn_kwargs=None, meth_kwargs=None):
-        if fn_kwargs is None:
-            fn_kwargs = {}
+    def _test_class_geom_op(self, method, meth_kwargs=None):
         if meth_kwargs is None:
             meth_kwargs = {}
+
         tensor, pil_img = self._create_data(height=10, width=10)
-        transformed_tensor = getattr(F, func)(tensor, **fn_kwargs)
-        transformed_pil_img = getattr(F, func)(pil_img, **fn_kwargs)
+        # test for class interface
+        f = getattr(T, method)(**meth_kwargs)
+        scripted_fn = torch.jit.script(f)
+
+        # set seed to reproduce the same transformation for tensor and PIL image
+        torch.manual_seed(12)
+        transformed_tensor = f(tensor)
+        torch.manual_seed(12)
+        transformed_pil_img = f(pil_img)
         self.compareTensorToPIL(transformed_tensor, transformed_pil_img)
 
-        scripted_fn = torch.jit.script(getattr(F, func))
-        transformed_tensor_script = scripted_fn(tensor, **fn_kwargs)
+        torch.manual_seed(12)
+        transformed_tensor_script = scripted_fn(tensor)
         self.assertTrue(transformed_tensor.equal(transformed_tensor_script))
 
-        # test for class interface
-        f = getattr(T, method)(**meth_kwargs)
-        scripted_fn = torch.jit.script(f)
-        scripted_fn(tensor)
+    def _test_geom_op(self, func, method, fn_kwargs=None, meth_kwargs=None):
+        self._test_functional_geom_op(func, fn_kwargs)
+        self._test_class_geom_op(method, meth_kwargs)
 
     def test_random_horizontal_flip(self):
         self._test_geom_op('hflip', 'RandomHorizontalFlip')
@@ -107,21 +112,20 @@ class Tester(unittest.TestCase):
             'crop', 'RandomCrop', fn_kwargs=fn_kwargs, meth_kwargs=meth_kwargs
         )
 
-        tensor = torch.randint(0, 255, (3, 10, 10), dtype=torch.uint8)
-        # Test torchscript of transforms.RandomCrop with size as int
-        f = T.RandomCrop(size=5)
-        scripted_fn = torch.jit.script(f)
-        scripted_fn(tensor)
-
-        # Test torchscript of transforms.RandomCrop with size as [int, ]
-        f = T.RandomCrop(size=[5, ], padding=[2, ])
-        scripted_fn = torch.jit.script(f)
-        scripted_fn(tensor)
-
-        # Test torchscript of transforms.RandomCrop with size as list
-        f = T.RandomCrop(size=[6, 6])
-        scripted_fn = torch.jit.script(f)
-        scripted_fn(tensor)
+        sizes = [5, [5, ], [6, 6]]
+        padding_configs = [
+            {"padding_mode": "constant", "fill": 0},
+            {"padding_mode": "constant", "fill": 10},
+            {"padding_mode": "constant", "fill": 20},
+            {"padding_mode": "edge"},
+            {"padding_mode": "reflect"},
+        ]
+
+        for size in sizes:
+            for padding_config in padding_configs:
+                config = dict(padding_config)
+                config["size"] = size
+                self._test_class_geom_op("RandomCrop", config)
 
     def test_center_crop(self):
         fn_kwargs = {"output_size": (4, 5)}

torchvision/transforms/functional.py

@@ -371,7 +371,7 @@ def pad(img: Tensor, padding: List[int], fill: int = 0, padding_mode: str = "con
             length 3, it is used to fill R, G, B channels respectively.
             This value is only used when the padding_mode is constant. Only int value is supported for Tensors.
         padding_mode: Type of padding. Should be: constant, edge, reflect or symmetric. Default is constant.
-            Only "constant" is supported for Tensors as of now.
+            Mode symmetric is not yet supported for Tensor inputs.
 
             - constant: pads with a constant value, this value is specified with fill
 

torchvision/transforms/functional_tensor.py

@@ -368,7 +368,8 @@ def pad(img: Tensor, padding: List[int], fill: int = 0, padding_mode: str = "con
             list of length 1: ``[padding, ]``.
         fill (int): Pixel fill value for constant fill. Default is 0.
             This value is only used when the padding_mode is constant
-        padding_mode (str): Type of padding. Only "constant" is supported for Tensors as of now.
+        padding_mode (str): Type of padding. Should be: constant, edge or reflect. Default is constant.
+            Mode symmetric is not yet supported for Tensor inputs.
 
             - constant: pads with a constant value, this value is specified with fill
 

torchvision/transforms/transforms.py

@@ -305,7 +305,7 @@ class Pad(torch.nn.Module):
             length 3, it is used to fill R, G, B channels respectively.
             This value is only used when the padding_mode is constant
         padding_mode (str): Type of padding. Should be: constant, edge, reflect or symmetric.
-            Default is constant. Only "constant" is supported for Tensors as of now.
+            Default is constant. Mode symmetric is not yet supported for Tensor inputs.
 
             - constant: pads with a constant value, this value is specified with fill
 
@@ -469,6 +469,7 @@ class RandomCrop(torch.nn.Module):
             length 3, it is used to fill R, G, B channels respectively.
             This value is only used when the padding_mode is constant
         padding_mode (str): Type of padding. Should be: constant, edge, reflect or symmetric. Default is constant.
+            Mode symmetric is not yet supported for Tensor inputs.
 
              - constant: pads with a constant value, this value is specified with fill