[fbsync] Prototype transforms cleanup (#5504)

Summary:
* fix grayscale to RGB for batches

* make unsupported types in auto augment a parameter

* make auto augment kwargs explicit

* add missing error message

* add support for specifying probabilites on RandomChoice

* remove TODO for deprecating p on random transforms

* streamline sample type checking

* address comments

* split image_size into height and width in auto augment

Reviewed By: datumbox

Differential Revision: D34579511

fbshipit-source-id: 757663a5a77f229cd1592b4c23dc17c7e8fe4807

Co-authored-by: Vasilis Vryniotis <[email protected]>

Author: 2 people

torchvision/prototype/transforms/_augment.py

@@ -7,7 +7,7 @@
 from torchvision.prototype import features
 from torchvision.prototype.transforms import Transform, functional as F
 
-from ._utils import query_image, get_image_dimensions
+from ._utils import query_image, get_image_dimensions, has_all, has_any
 
 
 class RandomErasing(Transform):
@@ -33,7 +33,6 @@ def __init__(
             raise ValueError("Scale should be between 0 and 1")
         if p < 0 or p > 1:
             raise ValueError("Random erasing probability should be between 0 and 1")
-        # TODO: deprecate p in favor of wrapping the transform in a RandomApply
         self.p = p
         self.scale = scale
         self.ratio = ratio
@@ -88,9 +87,7 @@ def _get_params(self, sample: Any) -> Dict[str, Any]:
         return dict(zip("ijhwv", (i, j, h, w, v)))
 
     def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
-        if isinstance(input, (features.BoundingBox, features.SegmentationMask)):
-            raise TypeError(f"{type(input).__name__}'s are not supported by {type(self).__name__}()")
-        elif isinstance(input, features.Image):
+        if isinstance(input, features.Image):
             output = F.erase_image_tensor(input, **params)
             return features.Image.new_like(input, output)
         elif isinstance(input, torch.Tensor):
@@ -99,10 +96,13 @@ def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
             return input
 
     def forward(self, *inputs: Any) -> Any:
-        if torch.rand(1) >= self.p:
-            return inputs if len(inputs) > 1 else inputs[0]
+        sample = inputs if len(inputs) > 1 else inputs[0]
+        if has_any(sample, features.BoundingBox, features.SegmentationMask):
+            raise TypeError(f"BoundingBox'es and SegmentationMask's are not supported by {type(self).__name__}()")
+        elif torch.rand(1) >= self.p:
+            return sample
 
-        return super().forward(*inputs)
+        return super().forward(sample)
 
 
 class RandomMixup(Transform):
@@ -115,9 +115,7 @@ def _get_params(self, sample: Any) -> Dict[str, Any]:
         return dict(lam=float(self._dist.sample(())))
 
     def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
-        if isinstance(input, (features.BoundingBox, features.SegmentationMask)):
-            raise TypeError(f"{type(input).__name__}'s are not supported by {type(self).__name__}()")
-        elif isinstance(input, features.Image):
+        if isinstance(input, features.Image):
             output = F.mixup_image_tensor(input, **params)
             return features.Image.new_like(input, output)
         elif isinstance(input, features.OneHotLabel):
@@ -126,6 +124,14 @@ def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
         else:
             return input
 
+    def forward(self, *inputs: Any) -> Any:
+        sample = inputs if len(inputs) > 1 else inputs[0]
+        if has_any(sample, features.BoundingBox, features.SegmentationMask):
+            raise TypeError(f"BoundingBox'es and SegmentationMask's are not supported by {type(self).__name__}()")
+        elif not has_all(sample, features.Image, features.OneHotLabel):
+            raise TypeError(f"{type(self).__name__}() is only defined for Image's *and* OneHotLabel's.")
+        return super().forward(sample)
+
 
 class RandomCutmix(Transform):
     def __init__(self, *, alpha: float) -> None:
@@ -157,13 +163,19 @@ def _get_params(self, sample: Any) -> Dict[str, Any]:
         return dict(box=box, lam_adjusted=lam_adjusted)
 
     def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
-        if isinstance(input, (features.BoundingBox, features.SegmentationMask)):
-            raise TypeError(f"{type(input).__name__}'s are not supported by {type(self).__name__}()")
-        elif isinstance(input, features.Image):
+        if isinstance(input, features.Image):
             output = F.cutmix_image_tensor(input, box=params["box"])
             return features.Image.new_like(input, output)
         elif isinstance(input, features.OneHotLabel):
             output = F.cutmix_one_hot_label(input, lam_adjusted=params["lam_adjusted"])
             return features.OneHotLabel.new_like(input, output)
         else:
             return input
+
+    def forward(self, *inputs: Any) -> Any:
+        sample = inputs if len(inputs) > 1 else inputs[0]
+        if has_any(sample, features.BoundingBox, features.SegmentationMask):
+            raise TypeError(f"BoundingBox'es and SegmentationMask's are not supported by {type(self).__name__}()")
+        elif not has_all(sample, features.Image, features.OneHotLabel):
+            raise TypeError(f"{type(self).__name__}() is only defined for Image's *and* OneHotLabel's.")
+        return super().forward(sample)