[fbsync] [prototype] Gaussian Blur clean up (#6888)

Summary:
* Refactor gaussian_blur

* Add conditional reshape

* Further refactoring

* Remove unused import.

Reviewed By: datumbox

Differential Revision: D41020542

fbshipit-source-id: 72694024272d91818c4154f7b5f7097e6d21154f

Author: NicolasHug

torchvision/prototype/transforms/functional/_misc.py

@@ -5,7 +5,6 @@
 import torch
 from torch.nn.functional import conv2d, pad as torch_pad
 from torchvision.prototype import features
-from torchvision.transforms import functional_tensor as _FT
 from torchvision.transforms.functional import pil_to_tensor, to_pil_image
 
 
@@ -68,9 +67,9 @@ def normalize(
 
 
 def _get_gaussian_kernel1d(kernel_size: int, sigma: float, dtype: torch.dtype, device: torch.device) -> torch.Tensor:
-    lim = (kernel_size - 1) / (2 * math.sqrt(2) * sigma)
+    lim = (kernel_size - 1) / (2.0 * math.sqrt(2.0) * sigma)
     x = torch.linspace(-lim, lim, steps=kernel_size, dtype=dtype, device=device)
-    kernel1d = torch.softmax(-x.pow_(2), dim=0)
+    kernel1d = torch.softmax(x.pow_(2).neg_(), dim=0)
     return kernel1d
 
 
@@ -89,54 +88,61 @@ def gaussian_blur_image_tensor(
     # TODO: consider deprecating integers from sigma on the future
     if isinstance(kernel_size, int):
         kernel_size = [kernel_size, kernel_size]
-    if len(kernel_size) != 2:
+    elif len(kernel_size) != 2:
         raise ValueError(f"If kernel_size is a sequence its length should be 2. Got {len(kernel_size)}")
     for ksize in kernel_size:
         if ksize % 2 == 0 or ksize < 0:
             raise ValueError(f"kernel_size should have odd and positive integers. Got {kernel_size}")
 
     if sigma is None:
         sigma = [ksize * 0.15 + 0.35 for ksize in kernel_size]
-
-    if sigma is not None and not isinstance(sigma, (int, float, list, tuple)):
-        raise TypeError(f"sigma should be either float or sequence of floats. Got {type(sigma)}")
-    if isinstance(sigma, (int, float)):
-        sigma = [float(sigma), float(sigma)]
-    if isinstance(sigma, (list, tuple)) and len(sigma) == 1:
-        sigma = [sigma[0], sigma[0]]
-    if len(sigma) != 2:
-        raise ValueError(f"If sigma is a sequence, its length should be 2. Got {len(sigma)}")
+    else:
+        if isinstance(sigma, (list, tuple)):
+            length = len(sigma)
+            if length == 1:
+                s = float(sigma[0])
+                sigma = [s, s]
+            elif length != 2:
+                raise ValueError(f"If sigma is a sequence, its length should be 2. Got {length}")
+        elif isinstance(sigma, (int, float)):
+            s = float(sigma)
+            sigma = [s, s]
+        else:
+            raise TypeError(f"sigma should be either float or sequence of floats. Got {type(sigma)}")
     for s in sigma:
         if s <= 0.0:
             raise ValueError(f"sigma should have positive values. Got {sigma}")
 
     if image.numel() == 0:
         return image
 
+    dtype = image.dtype
     shape = image.shape
-
-    if image.ndim > 4:
+    ndim = image.ndim
+    if ndim == 3:
+        image = image.unsqueeze(dim=0)
+    elif ndim > 4:
         image = image.reshape((-1,) + shape[-3:])
-        needs_unsquash = True
-    else:
-        needs_unsquash = False
 
-    dtype = image.dtype if torch.is_floating_point(image) else torch.float32
-    kernel = _get_gaussian_kernel2d(kernel_size, sigma, dtype=dtype, device=image.device)
-    kernel = kernel.expand(image.shape[-3], 1, kernel.shape[0], kernel.shape[1])
+    fp = torch.is_floating_point(image)
+    kernel = _get_gaussian_kernel2d(kernel_size, sigma, dtype=dtype if fp else torch.float32, device=image.device)
+    kernel = kernel.expand(shape[-3], 1, kernel.shape[0], kernel.shape[1])
 
-    image, need_cast, need_squeeze, out_dtype = _FT._cast_squeeze_in(image, [kernel.dtype])
+    output = image if fp else image.to(dtype=torch.float32)
 
     # padding = (left, right, top, bottom)
     padding = [kernel_size[0] // 2, kernel_size[0] // 2, kernel_size[1] // 2, kernel_size[1] // 2]
-    output = torch_pad(image, padding, mode="reflect")
-    output = conv2d(output, kernel, groups=output.shape[-3])
-
-    output = _FT._cast_squeeze_out(output, need_cast, need_squeeze, out_dtype)
+    output = torch_pad(output, padding, mode="reflect")
+    output = conv2d(output, kernel, groups=shape[-3])
 
-    if needs_unsquash:
+    if ndim == 3:
+        output = output.squeeze(dim=0)
+    elif ndim > 4:
         output = output.reshape(shape)
 
+    if not fp:
+        output = output.round_().to(dtype=dtype)
+
     return output