From 286e3166cb61e45038245b979db31aac60d1745b Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 10 Apr 2020 10:16:07 -0400
Subject: [PATCH 01/23] Adds as_tensor to functional.py

Similar functionality to to_tensor without the default conversion to float and division by 255.
Also adds support for Image mode 'L'.
---
 torchvision/transforms/functional.py | 47 ++++++++++++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/torchvision/transforms/functional.py b/torchvision/transforms/functional.py
index 7ce1fb6ab36..746ca9eed77 100644
--- a/torchvision/transforms/functional.py
+++ b/torchvision/transforms/functional.py
@@ -81,6 +81,53 @@ def to_tensor(pic):
     else:
         return img
 
+def as_tensor(pic):
+    """Convert a ``PIL Image`` or ``numpy.ndarray`` to tensor of same type.
+
+    See ``AsTensor`` for more details.
+
+    Args:
+        pic (PIL Image or numpy.ndarray): Image to be converted to tensor.
+
+    Returns:
+        Tensor: Converted image.
+    """
+    if not(_is_pil_image(pic) or _is_numpy(pic)):
+        raise TypeError('pic should be PIL Image or ndarray. Got {}'.format(type(pic)))
+
+    if _is_numpy(pic) and not _is_numpy_image(pic):
+        raise ValueError('pic should be 2/3 dimensional. Got {} dimensions.'.format(pic.ndim))
+
+    if isinstance(pic, np.ndarray):
+        # handle numpy array
+        if pic.ndim == 2:
+            pic = pic[:, :, None]
+
+        img = torch.from_numpy(pic.transpose((2, 0, 1)))
+        return img
+
+    if accimage is not None and isinstance(pic, accimage.Image):
+        nppic = np.zeros([pic.channels, pic.height, pic.width], dtype=np.float32)
+        pic.copyto(nppic)
+        return torch.from_numpy(nppic)
+
+    # handle PIL Image
+    if pic.mode == 'I':
+        img = torch.from_numpy(np.array(pic, np.int32, copy=False))
+    elif pic.mode == 'I;16':
+        img = torch.from_numpy(np.array(pic, np.int16, copy=False))
+    elif pic.mode == 'F':
+        img = torch.from_numpy(np.array(pic, np.float32, copy=False))
+    elif pic.mode == '1' or pic.mode == 'L':
+        img = torch.from_numpy(np.array(pic, np.uint8, copy=False))
+    else:
+        img = torch.ByteTensor(torch.ByteStorage.from_buffer(pic.tobytes()))
+
+    img = img.view(pic.size[1], pic.size[0], len(pic.getbands()))
+    # put it from HWC to CHW format
+    img = img.permute((2, 0, 1)).contiguous()
+    return img
+
 
 def to_pil_image(pic, mode=None):
     """Convert a tensor or an ndarray to PIL Image.

From f7eb48997520780c8dcffab5bc5e8c6ea87c2e2b Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 10 Apr 2020 10:42:38 -0400
Subject: [PATCH 02/23] Adds tests to AsTensor()

Adds tests to AsTensor and removes the conversion to float and division by 255.
---
 test/test_transforms.py | 44 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index f19c5480b02..f2c9b71f500 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -512,6 +512,50 @@ def test_accimage_to_tensor(self):
         self.assertEqual(expected_output.size(), output.size())
         self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
 
+    def test_as_tensor(self):
+        test_channels = [1, 3, 4]
+        height, width = 4, 4
+        trans = transforms.AsTensor()
+
+        with self.assertRaises(TypeError):
+            trans(np.random.rand(1, height, width).tolist())
+
+        with self.assertRaises(ValueError):
+            trans(np.random.rand(height))
+            trans(np.random.rand(1, 1, height, width))
+
+        for channels in test_channels:
+            input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
+            img = transforms.ToPILImage()(input_data)
+            output = trans(img)
+            self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
+
+            ndarray = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)
+            output = trans(ndarray)
+            expected_output = ndarray.transpose((2, 0, 1))
+            self.assertTrue(np.allclose(output.numpy(), expected_output))
+
+            ndarray = np.random.rand(height, width, channels).astype(np.float32)
+            output = trans(ndarray)
+            expected_output = ndarray.transpose((2, 0, 1))
+            self.assertTrue(np.allclose(output.numpy(), expected_output))
+
+        # separate test for mode '1' PIL images
+        input_data = torch.ByteTensor(1, height, width).bernoulli_()
+        img = transforms.ToPILImage()(input_data.mul(255)).convert('1')
+        output = trans(img)
+        self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
+
+    @unittest.skipIf(accimage is None, 'accimage not available')
+    def test_accimage_as_tensor(self):
+        trans = transforms.AsTensor()
+
+        expected_output = trans(Image.open(GRACE_HOPPER).convert('RGB'))
+        output = trans(accimage.Image(GRACE_HOPPER))
+
+        self.assertEqual(expected_output.size(), output.size())
+        self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
+
     @unittest.skipIf(accimage is None, 'accimage not available')
     def test_accimage_resize(self):
         trans = transforms.Compose([

From f90b3bcb45d5cc614702f404c1ba64c6354063a5 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 10 Apr 2020 10:47:59 -0400
Subject: [PATCH 03/23] Adds AsTensor to transforms.py

Calls the as_tensor function in functionals and adds the function AsTensor as callable from transforms.
---
 torchvision/transforms/transforms.py | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/torchvision/transforms/transforms.py b/torchvision/transforms/transforms.py
index 10783c8e53d..a7c33bf3691 100644
--- a/torchvision/transforms/transforms.py
+++ b/torchvision/transforms/transforms.py
@@ -15,7 +15,7 @@
 from . import functional as F
 
 
-__all__ = ["Compose", "ToTensor", "ToPILImage", "Normalize", "Resize", "Scale", "CenterCrop", "Pad",
+__all__ = ["Compose", "ToTensor", "AsTensor", "ToPILImage", "Normalize", "Resize", "Scale", "CenterCrop", "Pad",
            "Lambda", "RandomApply", "RandomChoice", "RandomOrder", "RandomCrop", "RandomHorizontalFlip",
            "RandomVerticalFlip", "RandomResizedCrop", "RandomSizedCrop", "FiveCrop", "TenCrop", "LinearTransformation",
            "ColorJitter", "RandomRotation", "RandomAffine", "Grayscale", "RandomGrayscale",
@@ -95,6 +95,28 @@ def __repr__(self):
         return self.__class__.__name__ + '()'
 
 
+class AsTensor(object):
+    """Convert a ``PIL Image`` or ``numpy.ndarray`` to tensor of the same type.
+
+    Converts a PIL Image or numpy.ndarray (H x W x C) to a torch.Tensor of shape (C x H x W)
+    if the PIL Image belongs to one of the modes (L, LA, P, I, F, RGB, YCbCr, RGBA, CMYK, 1)
+    or if the numpy.ndarray has dtype = np.uint8
+    """
+
+    def __call__(self, pic):
+        """
+        Args:
+            pic (PIL Image or numpy.ndarray): Image to be converted to tensor.
+
+        Returns:
+            Tensor: Converted image.
+        """
+        return F.as_tensor(pic)
+
+    def __repr__(self):
+        return self.__class__.__name__ + '()'
+
+
 class ToPILImage(object):
     """Convert a tensor or an ndarray to PIL Image.
 

From 9c2fd3b465ea16033b0b080672612010c319a8c7 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 10 Apr 2020 10:49:36 -0400
Subject: [PATCH 04/23] Removes the pic.mode == 'L'

This was handled by the else condition previously so I'll remove it.
---
 torchvision/transforms/functional.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/torchvision/transforms/functional.py b/torchvision/transforms/functional.py
index 746ca9eed77..53e2420a6e3 100644
--- a/torchvision/transforms/functional.py
+++ b/torchvision/transforms/functional.py
@@ -118,7 +118,7 @@ def as_tensor(pic):
         img = torch.from_numpy(np.array(pic, np.int16, copy=False))
     elif pic.mode == 'F':
         img = torch.from_numpy(np.array(pic, np.float32, copy=False))
-    elif pic.mode == '1' or pic.mode == 'L':
+    elif pic.mode == '1':
         img = torch.from_numpy(np.array(pic, np.uint8, copy=False))
     else:
         img = torch.ByteTensor(torch.ByteStorage.from_buffer(pic.tobytes()))

From 08ab5ecf3516ffb89e06b90db343d163922d4931 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 10 Apr 2020 11:59:59 -0400
Subject: [PATCH 05/23] Fix Lint issue

Adds two line breaks between functions to fix lint issue
---
 torchvision/transforms/functional.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/torchvision/transforms/functional.py b/torchvision/transforms/functional.py
index 53e2420a6e3..0048507e572 100644
--- a/torchvision/transforms/functional.py
+++ b/torchvision/transforms/functional.py
@@ -81,6 +81,7 @@ def to_tensor(pic):
     else:
         return img
 
+
 def as_tensor(pic):
     """Convert a ``PIL Image`` or ``numpy.ndarray`` to tensor of same type.
 

From cb19ed41c0cb19c56fce8ffbd32aa19a462cfdbc Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Wed, 15 Apr 2020 22:49:34 -0400
Subject: [PATCH 06/23] Replace from_numpy with as_tensor

Removes the extra if conditionals and replaces from_numpy with as_tensor.
---
 torchvision/transforms/functional.py | 17 ++++-------------
 1 file changed, 4 insertions(+), 13 deletions(-)

diff --git a/torchvision/transforms/functional.py b/torchvision/transforms/functional.py
index 0048507e572..cd1696bf7fb 100644
--- a/torchvision/transforms/functional.py
+++ b/torchvision/transforms/functional.py
@@ -104,29 +104,20 @@ def as_tensor(pic):
         if pic.ndim == 2:
             pic = pic[:, :, None]
 
-        img = torch.from_numpy(pic.transpose((2, 0, 1)))
+        img = torch.as_tensor(pic.transpose((2, 0, 1)))
         return img
 
     if accimage is not None and isinstance(pic, accimage.Image):
         nppic = np.zeros([pic.channels, pic.height, pic.width], dtype=np.float32)
         pic.copyto(nppic)
-        return torch.from_numpy(nppic)
+        return torch.as_tensor(nppic)
 
     # handle PIL Image
-    if pic.mode == 'I':
-        img = torch.from_numpy(np.array(pic, np.int32, copy=False))
-    elif pic.mode == 'I;16':
-        img = torch.from_numpy(np.array(pic, np.int16, copy=False))
-    elif pic.mode == 'F':
-        img = torch.from_numpy(np.array(pic, np.float32, copy=False))
-    elif pic.mode == '1':
-        img = torch.from_numpy(np.array(pic, np.uint8, copy=False))
-    else:
-        img = torch.ByteTensor(torch.ByteStorage.from_buffer(pic.tobytes()))
+    img = torch.as_tensor(np.asarray(pic))
 
     img = img.view(pic.size[1], pic.size[0], len(pic.getbands()))
     # put it from HWC to CHW format
-    img = img.permute((2, 0, 1)).contiguous()
+    img = img.permute((2, 0, 1))
     return img
 
 

From 38ad5f3986d6d9c90903ff11c044065a8649dae8 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 16:45:55 -0400
Subject: [PATCH 07/23] Renames as_tensor to pil_to_tensor

Renames the function as_tensor to pil_to_tensor and narrows the scope of the function.  At the same time also creates a flag that defaults to True for swapping to the channels first format.
---
 torchvision/transforms/functional.py | 29 ++++++++++------------------
 1 file changed, 10 insertions(+), 19 deletions(-)

diff --git a/torchvision/transforms/functional.py b/torchvision/transforms/functional.py
index cd1696bf7fb..21a0a32a2cf 100644
--- a/torchvision/transforms/functional.py
+++ b/torchvision/transforms/functional.py
@@ -82,30 +82,20 @@ def to_tensor(pic):
         return img
 
 
-def as_tensor(pic):
-    """Convert a ``PIL Image`` or ``numpy.ndarray`` to tensor of same type.
+def pil_to_tensor(pic, swap_to_channelsfirst=True):
+    """Convert a ``PIL Image`` to a tensor of the same type.
 
     See ``AsTensor`` for more details.
 
     Args:
-        pic (PIL Image or numpy.ndarray): Image to be converted to tensor.
+        pic (PIL Image): Image to be converted to tensor.
+        swap_to_channelsfirst (bool): Boolean indicator to convert to CHW format.
 
     Returns:
         Tensor: Converted image.
     """
-    if not(_is_pil_image(pic) or _is_numpy(pic)):
-        raise TypeError('pic should be PIL Image or ndarray. Got {}'.format(type(pic)))
-
-    if _is_numpy(pic) and not _is_numpy_image(pic):
-        raise ValueError('pic should be 2/3 dimensional. Got {} dimensions.'.format(pic.ndim))
-
-    if isinstance(pic, np.ndarray):
-        # handle numpy array
-        if pic.ndim == 2:
-            pic = pic[:, :, None]
-
-        img = torch.as_tensor(pic.transpose((2, 0, 1)))
-        return img
+    if not(_is_pil_image(pic)):
+        raise TypeError('pic should be PIL Image. Got {}'.format(type(pic)))
 
     if accimage is not None and isinstance(pic, accimage.Image):
         nppic = np.zeros([pic.channels, pic.height, pic.width], dtype=np.float32)
@@ -115,9 +105,10 @@ def as_tensor(pic):
     # handle PIL Image
     img = torch.as_tensor(np.asarray(pic))
 
-    img = img.view(pic.size[1], pic.size[0], len(pic.getbands()))
-    # put it from HWC to CHW format
-    img = img.permute((2, 0, 1))
+    if swap_to_channelsfirst:
+        img = img.view(pic.size[1], pic.size[0], len(pic.getbands()))
+        # put it from HWC to CHW format
+        img = img.permute((2, 0, 1))
     return img
 
 

From 1fa91a844075a4b4e7da9568615645a477e2785b Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 17:04:52 -0400
Subject: [PATCH 08/23] Renames AsTensor to PILToImage

Renames the function AsTensor to PILToImage and modifies the description.  Adds the swap_to_channelsfirst boolean variable to indicate if the user wishes to change the shape of the input.
---
 torchvision/transforms/transforms.py | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/torchvision/transforms/transforms.py b/torchvision/transforms/transforms.py
index a7c33bf3691..ee5fd6c3218 100644
--- a/torchvision/transforms/transforms.py
+++ b/torchvision/transforms/transforms.py
@@ -15,7 +15,7 @@
 from . import functional as F
 
 
-__all__ = ["Compose", "ToTensor", "AsTensor", "ToPILImage", "Normalize", "Resize", "Scale", "CenterCrop", "Pad",
+__all__ = ["Compose", "ToTensor", "PILToTensor", "ToPILImage", "Normalize", "Resize", "Scale", "CenterCrop", "Pad",
            "Lambda", "RandomApply", "RandomChoice", "RandomOrder", "RandomCrop", "RandomHorizontalFlip",
            "RandomVerticalFlip", "RandomResizedCrop", "RandomSizedCrop", "FiveCrop", "TenCrop", "LinearTransformation",
            "ColorJitter", "RandomRotation", "RandomAffine", "Grayscale", "RandomGrayscale",
@@ -95,23 +95,24 @@ def __repr__(self):
         return self.__class__.__name__ + '()'
 
 
-class AsTensor(object):
-    """Convert a ``PIL Image`` or ``numpy.ndarray`` to tensor of the same type.
+class PILToTensor(object):
+    """Convert a ``PIL Image`` to a tensor of the same type.
 
-    Converts a PIL Image or numpy.ndarray (H x W x C) to a torch.Tensor of shape (C x H x W)
-    if the PIL Image belongs to one of the modes (L, LA, P, I, F, RGB, YCbCr, RGBA, CMYK, 1)
-    or if the numpy.ndarray has dtype = np.uint8
+    Converts a PIL Image (H x W x C) to a torch.Tensor. If swap_to_channelsfirst is True
+    the returned shape will be (C x H x W) otherwise the shape will remain unchanged.
     """
 
-    def __call__(self, pic):
+    def __call__(self, pic, swap_to_channelsfirst=True):
         """
         Args:
-            pic (PIL Image or numpy.ndarray): Image to be converted to tensor.
+            pic (PIL Image): Image to be converted to tensor.
+            swap_to_channelsfirst (bool): Boolean indicator to swap to channels first format. 
+                Defaults to True.
 
         Returns:
             Tensor: Converted image.
         """
-        return F.as_tensor(pic)
+        return F.pil_to_tensor(pic)
 
     def __repr__(self):
         return self.__class__.__name__ + '()'

From 0fefbcbd72e24dee2b8ab845434e8e570ce87e31 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 17:30:05 -0400
Subject: [PATCH 09/23] Add the __init__ function to PILToTensor

Add the __init__ function to PILToTensor since it contains the swap_to_channelsfirst parameter now.
---
 torchvision/transforms/transforms.py | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/torchvision/transforms/transforms.py b/torchvision/transforms/transforms.py
index ee5fd6c3218..653f9228d29 100644
--- a/torchvision/transforms/transforms.py
+++ b/torchvision/transforms/transforms.py
@@ -101,8 +101,10 @@ class PILToTensor(object):
     Converts a PIL Image (H x W x C) to a torch.Tensor. If swap_to_channelsfirst is True
     the returned shape will be (C x H x W) otherwise the shape will remain unchanged.
     """
+    def __init__(self, swap_to_channelsfirst=True):
+        self.swap_to_channelsfirst = swap_to_channelsfirst
 
-    def __call__(self, pic, swap_to_channelsfirst=True):
+    def __call__(self, pic):
         """
         Args:
             pic (PIL Image): Image to be converted to tensor.
@@ -112,7 +114,7 @@ def __call__(self, pic, swap_to_channelsfirst=True):
         Returns:
             Tensor: Converted image.
         """
-        return F.pil_to_tensor(pic)
+        return F.pil_to_tensor(pic, self.swap_to_channelsfirst)
 
     def __repr__(self):
         return self.__class__.__name__ + '()'

From 7662b23f0e974dfbab3c950cb0cc1588af304ef3 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 17:32:26 -0400
Subject: [PATCH 10/23] fix lint issue

remove trailing white space
---
 torchvision/transforms/transforms.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/torchvision/transforms/transforms.py b/torchvision/transforms/transforms.py
index 653f9228d29..4699d71515e 100644
--- a/torchvision/transforms/transforms.py
+++ b/torchvision/transforms/transforms.py
@@ -108,7 +108,7 @@ def __call__(self, pic):
         """
         Args:
             pic (PIL Image): Image to be converted to tensor.
-            swap_to_channelsfirst (bool): Boolean indicator to swap to channels first format. 
+            swap_to_channelsfirst (bool): Boolean indicator to swap to channels first format.
                 Defaults to True.
 
         Returns:

From 75be7bbfac9f2589fb03f222e1a0945e00e6abff Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 17:37:24 -0400
Subject: [PATCH 11/23] Fix the tests

Reflects the name change to PILToTensor and the parameter to the function as well as the new narrowed scope that the function only accepts PIL images.
---
 test/test_transforms.py | 36 ++++++++++++++++++++++--------------
 1 file changed, 22 insertions(+), 14 deletions(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index f2c9b71f500..bcdb99be90f 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -512,32 +512,40 @@ def test_accimage_to_tensor(self):
         self.assertEqual(expected_output.size(), output.size())
         self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
 
-    def test_as_tensor(self):
+    def test_pil_to_tensor(self):
         test_channels = [1, 3, 4]
         height, width = 4, 4
-        trans = transforms.AsTensor()
+        trans = transforms.PILToTensor()
+        trans_noswap = transforms.PILToTensor(swap_to_channelsfirst=False)
 
         with self.assertRaises(TypeError):
             trans(np.random.rand(1, height, width).tolist())
-
-        with self.assertRaises(ValueError):
-            trans(np.random.rand(height))
-            trans(np.random.rand(1, 1, height, width))
+            trans(np.random.rand(1, height, width))
 
         for channels in test_channels:
             input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
             img = transforms.ToPILImage()(input_data)
             output = trans(img)
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
+            output = trans_noswap(img)
+            self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
-            ndarray = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)
-            output = trans(ndarray)
-            expected_output = ndarray.transpose((2, 0, 1))
+            input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)
+            img = transforms.ToPILImage()(input_data)
+            output = trans(img)
+            expected_output = input_data.transpose((2, 0, 1))
+            self.assertTrue(np.allclose(output.numpy(), expected_output))
+            output = trans_noswap(img)
+            expected_output = input_data
             self.assertTrue(np.allclose(output.numpy(), expected_output))
 
-            ndarray = np.random.rand(height, width, channels).astype(np.float32)
-            output = trans(ndarray)
-            expected_output = ndarray.transpose((2, 0, 1))
+            input_data = np.random.rand(height, width, channels).astype(np.float32)
+            img = transforms.ToPILImage()(input_data)
+            output = trans(img)
+            expected_output = input_data.transpose((2, 0, 1))
+            self.assertTrue(np.allclose(output.numpy(), expected_output))
+            output = trans_noswap(img)
+            expected_output = input_data
             self.assertTrue(np.allclose(output.numpy(), expected_output))
 
         # separate test for mode '1' PIL images
@@ -547,8 +555,8 @@ def test_as_tensor(self):
         self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
     @unittest.skipIf(accimage is None, 'accimage not available')
-    def test_accimage_as_tensor(self):
-        trans = transforms.AsTensor()
+    def test_accimage_pil_to_tensor(self):
+        trans = transforms.PILToTensor()
 
         expected_output = trans(Image.open(GRACE_HOPPER).convert('RGB'))
         output = trans(accimage.Image(GRACE_HOPPER))

From 123503acb65b6ebf2358500665df15833bf1ed00 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 17:54:55 -0400
Subject: [PATCH 12/23] fix tests

Instead of undoing the transpose just create a new tensor and test that one.
---
 test/test_transforms.py | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index bcdb99be90f..fd681fe4cde 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -527,6 +527,9 @@ def test_pil_to_tensor(self):
             img = transforms.ToPILImage()(input_data)
             output = trans(img)
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
+
+            input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
+            img = transforms.ToPILImage()(input_data)
             output = trans_noswap(img)
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
@@ -535,6 +538,9 @@ def test_pil_to_tensor(self):
             output = trans(img)
             expected_output = input_data.transpose((2, 0, 1))
             self.assertTrue(np.allclose(output.numpy(), expected_output))
+
+            input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)
+            img = transforms.ToPILImage()(input_data)
             output = trans_noswap(img)
             expected_output = input_data
             self.assertTrue(np.allclose(output.numpy(), expected_output))
@@ -544,6 +550,10 @@ def test_pil_to_tensor(self):
             output = trans(img)
             expected_output = input_data.transpose((2, 0, 1))
             self.assertTrue(np.allclose(output.numpy(), expected_output))
+            input_data = np.random.rand(height, width, channels).astype(np.float32)
+
+            input_data = np.random.rand(height, width, channels).astype(np.float32)
+            img = transforms.ToPILImage()(input_data)
             output = trans_noswap(img)
             expected_output = input_data
             self.assertTrue(np.allclose(output.numpy(), expected_output))

From eff1db0954867be1df7f061119bc7b7f12de43bf Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 19:46:54 -0400
Subject: [PATCH 13/23] Add the view back

Add img.view(pic.size[1], pic.size[0], len(pic.getbands())) back to outside the if condition.
---
 torchvision/transforms/functional.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/torchvision/transforms/functional.py b/torchvision/transforms/functional.py
index 21a0a32a2cf..ed48af7ef36 100644
--- a/torchvision/transforms/functional.py
+++ b/torchvision/transforms/functional.py
@@ -104,9 +104,9 @@ def pil_to_tensor(pic, swap_to_channelsfirst=True):
 
     # handle PIL Image
     img = torch.as_tensor(np.asarray(pic))
+    img = img.view(pic.size[1], pic.size[0], len(pic.getbands()))
 
     if swap_to_channelsfirst:
-        img = img.view(pic.size[1], pic.size[0], len(pic.getbands()))
         # put it from HWC to CHW format
         img = img.permute((2, 0, 1))
     return img

From 266860af078e27d57c5cabee34e02fdaecd2c2c9 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 20:09:41 -0400
Subject: [PATCH 14/23] fix test

fix conversion from torch tensor to PIL back to torch tensor.
---
 test/test_transforms.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index fd681fe4cde..bf525bee003 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -529,8 +529,8 @@ def test_pil_to_tensor(self):
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
             input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
-            img = transforms.ToPILImage()(input_data)
-            output = trans_noswap(img)
+            img = transforms.ToPILImage()(input_data) #HWC
+            output = trans_noswap(img).transpose((2, 0, 1)) #HWC -> CHW 
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
             input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)

From 610fc1e6dba9943b336e570547dc2fd1e44332a8 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 20:12:24 -0400
Subject: [PATCH 15/23] fix lint issues

---
 test/test_transforms.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index bf525bee003..fc9ca302941 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -529,8 +529,8 @@ def test_pil_to_tensor(self):
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
             input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
-            img = transforms.ToPILImage()(input_data) #HWC
-            output = trans_noswap(img).transpose((2, 0, 1)) #HWC -> CHW 
+            img = transforms.ToPILImage()(input_data)  # HWC
+            output = trans_noswap(img).transpose((2, 0, 1))  # HWC -> CHW 
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
             input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)

From b9cca779907e2e849fdc580d36cd78584e36f93a Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 20:14:09 -0400
Subject: [PATCH 16/23] fix lint

remove trailing white space
---
 test/test_transforms.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index fc9ca302941..a23196dc7ec 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -530,7 +530,7 @@ def test_pil_to_tensor(self):
 
             input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
             img = transforms.ToPILImage()(input_data)  # HWC
-            output = trans_noswap(img).transpose((2, 0, 1))  # HWC -> CHW 
+            output = trans_noswap(img).transpose((2, 0, 1))  # HWC -> CHW
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
             input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)

From 1b10f779dbf56159553a69b665884e1a6bf2fa2e Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 20:27:34 -0400
Subject: [PATCH 17/23] Fixed the channel swapping tensor test

Torch tranpose operates differently than numpy transpose.  Changed operation to permute.
---
 test/test_transforms.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index a23196dc7ec..787a211bde7 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -530,7 +530,7 @@ def test_pil_to_tensor(self):
 
             input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
             img = transforms.ToPILImage()(input_data)  # HWC
-            output = trans_noswap(img).transpose((2, 0, 1))  # HWC -> CHW
+            output = trans_noswap(img).permute(2, 0, 1)  # HWC -> CHW
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
             input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)

From fbf661c656a68d1826d3b24fac0395f3384d6301 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 20:42:46 -0400
Subject: [PATCH 18/23] Add mode='F'

Add mode information when converting to PIL Image from Float Tensor.
---
 test/test_transforms.py | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index 787a211bde7..fe227890db1 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -546,14 +546,13 @@ def test_pil_to_tensor(self):
             self.assertTrue(np.allclose(output.numpy(), expected_output))
 
             input_data = np.random.rand(height, width, channels).astype(np.float32)
-            img = transforms.ToPILImage()(input_data)
+            img = transforms.ToPILImage(mode='F')(input_data)
             output = trans(img)
             expected_output = input_data.transpose((2, 0, 1))
             self.assertTrue(np.allclose(output.numpy(), expected_output))
-            input_data = np.random.rand(height, width, channels).astype(np.float32)
 
             input_data = np.random.rand(height, width, channels).astype(np.float32)
-            img = transforms.ToPILImage()(input_data)
+            img = transforms.ToPILImage(mode='F')(input_data)
             output = trans_noswap(img)
             expected_output = input_data
             self.assertTrue(np.allclose(output.numpy(), expected_output))

From 598107f577eb646f2f3450fa8de7f396442715e5 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 21:46:18 -0400
Subject: [PATCH 19/23] Added inline comments to follow shape changes

---
 test/test_transforms.py | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index fe227890db1..ef87ddd84c7 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -529,7 +529,7 @@ def test_pil_to_tensor(self):
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
             input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
-            img = transforms.ToPILImage()(input_data)  # HWC
+            img = transforms.ToPILImage()(input_data)  # CHW -> HWC
             output = trans_noswap(img).permute(2, 0, 1)  # HWC -> CHW
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
@@ -545,17 +545,17 @@ def test_pil_to_tensor(self):
             expected_output = input_data
             self.assertTrue(np.allclose(output.numpy(), expected_output))
 
-            input_data = np.random.rand(height, width, channels).astype(np.float32)
-            img = transforms.ToPILImage(mode='F')(input_data)
-            output = trans(img)
-            expected_output = input_data.transpose((2, 0, 1))
-            self.assertTrue(np.allclose(output.numpy(), expected_output))
-
-            input_data = np.random.rand(height, width, channels).astype(np.float32)
-            img = transforms.ToPILImage(mode='F')(input_data)
-            output = trans_noswap(img)
+            input_data = torch.as_tensor(np.random.rand(channels, height, width).astype(np.float32))
+            img = transforms.ToPILImage()(input_data)  # CHW -> HWC
+            output = trans(img)  # HWC -> CHW
             expected_output = input_data
-            self.assertTrue(np.allclose(output.numpy(), expected_output))
+            self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
+
+            input_data = torch.as_tensor(np.random.rand(channels, height, width).astype(np.float32))
+            img = transforms.ToPILImage()(input_data)  # CHW -> HWC
+            output = trans_noswap(img)  # HWC -> HWC
+            expected_output = input_data.permute(1, 2, 0)  # CHW -> HWC
+            self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
 
         # separate test for mode '1' PIL images
         input_data = torch.ByteTensor(1, height, width).bernoulli_()

From d69048e15aad8a6c97292a1471f142ec2a9a61b6 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Thu, 16 Apr 2020 22:03:36 -0400
Subject: [PATCH 20/23] ToPILImage converts FloatTensors to uint8

---
 test/test_transforms.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index ef87ddd84c7..059873e5f29 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -546,15 +546,15 @@ def test_pil_to_tensor(self):
             self.assertTrue(np.allclose(output.numpy(), expected_output))
 
             input_data = torch.as_tensor(np.random.rand(channels, height, width).astype(np.float32))
-            img = transforms.ToPILImage()(input_data)  # CHW -> HWC
+            img = transforms.ToPILImage()(input_data)  # CHW -> HWC and (* 255).byte()
             output = trans(img)  # HWC -> CHW
-            expected_output = input_data
+            expected_output = (input_data * 255).byte()
             self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
 
             input_data = torch.as_tensor(np.random.rand(channels, height, width).astype(np.float32))
             img = transforms.ToPILImage()(input_data)  # CHW -> HWC
             output = trans_noswap(img)  # HWC -> HWC
-            expected_output = input_data.permute(1, 2, 0)  # CHW -> HWC
+            expected_output = (input_data.permute(1, 2, 0) * 255).byte()  # CHW -> HWC
             self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
 
         # separate test for mode '1' PIL images

From fa1084c950b77396835660218ab0c7109a68e0cd Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 15 May 2020 15:49:05 -0400
Subject: [PATCH 21/23] Remove testing not swapping

---
 test/test_transforms.py | 18 ------------------
 1 file changed, 18 deletions(-)

diff --git a/test/test_transforms.py b/test/test_transforms.py
index 059873e5f29..978ab823c95 100644
--- a/test/test_transforms.py
+++ b/test/test_transforms.py
@@ -516,7 +516,6 @@ def test_pil_to_tensor(self):
         test_channels = [1, 3, 4]
         height, width = 4, 4
         trans = transforms.PILToTensor()
-        trans_noswap = transforms.PILToTensor(swap_to_channelsfirst=False)
 
         with self.assertRaises(TypeError):
             trans(np.random.rand(1, height, width).tolist())
@@ -528,35 +527,18 @@ def test_pil_to_tensor(self):
             output = trans(img)
             self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
 
-            input_data = torch.ByteTensor(channels, height, width).random_(0, 255)
-            img = transforms.ToPILImage()(input_data)  # CHW -> HWC
-            output = trans_noswap(img).permute(2, 0, 1)  # HWC -> CHW
-            self.assertTrue(np.allclose(input_data.numpy(), output.numpy()))
-
             input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)
             img = transforms.ToPILImage()(input_data)
             output = trans(img)
             expected_output = input_data.transpose((2, 0, 1))
             self.assertTrue(np.allclose(output.numpy(), expected_output))
 
-            input_data = np.random.randint(low=0, high=255, size=(height, width, channels)).astype(np.uint8)
-            img = transforms.ToPILImage()(input_data)
-            output = trans_noswap(img)
-            expected_output = input_data
-            self.assertTrue(np.allclose(output.numpy(), expected_output))
-
             input_data = torch.as_tensor(np.random.rand(channels, height, width).astype(np.float32))
             img = transforms.ToPILImage()(input_data)  # CHW -> HWC and (* 255).byte()
             output = trans(img)  # HWC -> CHW
             expected_output = (input_data * 255).byte()
             self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
 
-            input_data = torch.as_tensor(np.random.rand(channels, height, width).astype(np.float32))
-            img = transforms.ToPILImage()(input_data)  # CHW -> HWC
-            output = trans_noswap(img)  # HWC -> HWC
-            expected_output = (input_data.permute(1, 2, 0) * 255).byte()  # CHW -> HWC
-            self.assertTrue(np.allclose(output.numpy(), expected_output.numpy()))
-
         # separate test for mode '1' PIL images
         input_data = torch.ByteTensor(1, height, width).bernoulli_()
         img = transforms.ToPILImage()(input_data.mul(255)).convert('1')

From 2cb7a4fe1a8694a396489fb9b189dc8735b97645 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 15 May 2020 15:53:09 -0400
Subject: [PATCH 22/23] Removes the swap_channelsfirst parameter

Makes the channel swapping the default behavior.
---
 torchvision/transforms/transforms.py | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

diff --git a/torchvision/transforms/transforms.py b/torchvision/transforms/transforms.py
index 4699d71515e..eb49b99be93 100644
--- a/torchvision/transforms/transforms.py
+++ b/torchvision/transforms/transforms.py
@@ -98,23 +98,18 @@ def __repr__(self):
 class PILToTensor(object):
     """Convert a ``PIL Image`` to a tensor of the same type.
 
-    Converts a PIL Image (H x W x C) to a torch.Tensor. If swap_to_channelsfirst is True
-    the returned shape will be (C x H x W) otherwise the shape will remain unchanged.
+    Converts a PIL Image (H x W x C) to a torch.Tensor of shape (C x H x W).
     """
-    def __init__(self, swap_to_channelsfirst=True):
-        self.swap_to_channelsfirst = swap_to_channelsfirst
 
     def __call__(self, pic):
         """
         Args:
             pic (PIL Image): Image to be converted to tensor.
-            swap_to_channelsfirst (bool): Boolean indicator to swap to channels first format.
-                Defaults to True.
 
         Returns:
             Tensor: Converted image.
         """
-        return F.pil_to_tensor(pic, self.swap_to_channelsfirst)
+        return F.pil_to_tensor(pic)
 
     def __repr__(self):
         return self.__class__.__name__ + '()'

From 3d565fd721fa25be831c0bf65604d7d181f7fd63 Mon Sep 17 00:00:00 2001
From: Steven Basart <xksteven@users.noreply.github.com>
Date: Fri, 15 May 2020 16:10:17 -0400
Subject: [PATCH 23/23] Remove the swap_channelsfirst argument

Remove the swap_channelsfirst argument and makes the swapping the default functionality.
---
 torchvision/transforms/functional.py | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/torchvision/transforms/functional.py b/torchvision/transforms/functional.py
index ed48af7ef36..bdfa6567a82 100644
--- a/torchvision/transforms/functional.py
+++ b/torchvision/transforms/functional.py
@@ -82,14 +82,13 @@ def to_tensor(pic):
         return img
 
 
-def pil_to_tensor(pic, swap_to_channelsfirst=True):
+def pil_to_tensor(pic):
     """Convert a ``PIL Image`` to a tensor of the same type.
 
     See ``AsTensor`` for more details.
 
     Args:
         pic (PIL Image): Image to be converted to tensor.
-        swap_to_channelsfirst (bool): Boolean indicator to convert to CHW format.
 
     Returns:
         Tensor: Converted image.
@@ -105,10 +104,8 @@ def pil_to_tensor(pic, swap_to_channelsfirst=True):
     # handle PIL Image
     img = torch.as_tensor(np.asarray(pic))
     img = img.view(pic.size[1], pic.size[0], len(pic.getbands()))
-
-    if swap_to_channelsfirst:
-        # put it from HWC to CHW format
-        img = img.permute((2, 0, 1))
+    # put it from HWC to CHW format
+    img = img.permute((2, 0, 1))
     return img