[fbsync] fix {convert_format, clamp}_bounding_box (#7229)

Reviewed By: vmoens

Differential Revision: D44416268

fbshipit-source-id: 664c723e512d08c8bb2e32e8efefbabbe65b0122

Author: NicolasHug

torchvision/prototype/transforms/functional/_meta.py

@@ -230,8 +230,10 @@ def convert_format_bounding_box(
     elif isinstance(inpt, datapoints.BoundingBox):
         if old_format is not None:
             raise ValueError("For bounding box datapoint inputs, `old_format` must not be passed.")
-        output = _convert_format_bounding_box(inpt, old_format=inpt.format, new_format=new_format, inplace=inplace)
-        return datapoints.BoundingBox.wrap_like(inpt, output)
+        output = _convert_format_bounding_box(
+            inpt.as_subclass(torch.Tensor), old_format=inpt.format, new_format=new_format, inplace=inplace
+        )
+        return datapoints.BoundingBox.wrap_like(inpt, output, format=new_format)
     else:
         raise TypeError(
             f"Input can either be a plain tensor or a bounding box datapoint, but got {type(inpt)} instead."
@@ -266,7 +268,7 @@ def clamp_bounding_box(
     elif isinstance(inpt, datapoints.BoundingBox):
         if format is not None or spatial_size is not None:
             raise ValueError("For bounding box datapoint inputs, `format` and `spatial_size` must not be passed.")
-        output = _clamp_bounding_box(inpt, format=inpt.format, spatial_size=inpt.spatial_size)
+        output = _clamp_bounding_box(inpt.as_subclass(torch.Tensor), format=inpt.format, spatial_size=inpt.spatial_size)
         return datapoints.BoundingBox.wrap_like(inpt, output)
     else:
         raise TypeError(