[fbsync] Update transforms docs sub-structure (#7291)

Summary: Co-authored-by: Philip Meier <[email protected]>

Reviewed By: vmoens

Differential Revision: D44416596

fbshipit-source-id: 995a5ab32054b30ff674160a451708e1d117abe9

Author: NicolasHug

docs/source/transforms.rst

@@ -14,11 +14,10 @@ transformations.
 This is useful if you have to build a more complex transformation pipeline
 (e.g. in the case of segmentation tasks).
 
-Most transformations accept both `PIL <https://pillow.readthedocs.io>`_
-images and tensor images, although some transformations are :ref:`PIL-only
-<transforms_pil_only>` and some are :ref:`tensor-only
-<transforms_tensor_only>`. The :ref:`conversion_transforms` may be used to
-convert to and from PIL images.
+Most transformations accept both `PIL <https://pillow.readthedocs.io>`_ images
+and tensor images, although some transformations are PIL-only and some are
+tensor-only. The :ref:`conversion_transforms` may be used to convert to and from
+PIL images, or for converting dtypes and ranges.
 
 The transformations that accept tensor images also accept batches of tensor
 images. A Tensor Image is a tensor with ``(C, H, W)`` shape, where ``C`` is a
@@ -70,8 +69,10 @@ The following examples illustrate the use of the available transforms:
     produce the same results.
 
 
-Scriptable transforms
----------------------
+Transforms scriptability
+------------------------
+
+.. TODO: Add note about v2 scriptability (in next PR)
 
 In order to script the transformations, please use ``torch.nn.Sequential`` instead of :class:`Compose`.
 
@@ -89,39 +90,36 @@ Make sure to use only scriptable transformations, i.e. that work with ``torch.Te
 For any custom transformations to be used with ``torch.jit.script``, they should be derived from ``torch.nn.Module``.
 
 
-Compositions of transforms
---------------------------
+Geometry
+--------
 
 .. autosummary::
     :toctree: generated/
     :template: class.rst
 
-    Compose
-
+    Resize
+    RandomCrop
+    RandomResizedCrop
+    CenterCrop
+    FiveCrop
+    TenCrop
+    Pad
+    RandomAffine
+    RandomPerspective
+    RandomRotation
+    RandomHorizontalFlip
+    RandomVerticalFlip
 
-Transforms on PIL Image and torch.\*Tensor
-------------------------------------------
+Color
+-----
 
 .. autosummary::
     :toctree: generated/
     :template: class.rst
 
-    CenterCrop
     ColorJitter
-    FiveCrop
     Grayscale
-    Pad
-    RandomAffine
-    RandomApply
-    RandomCrop
     RandomGrayscale
-    RandomHorizontalFlip
-    RandomPerspective
-    RandomResizedCrop
-    RandomRotation
-    RandomVerticalFlip
-    Resize
-    TenCrop
     GaussianBlur
     RandomInvert
     RandomPosterize
@@ -130,23 +128,20 @@ Transforms on PIL Image and torch.\*Tensor
     RandomAutocontrast
     RandomEqualize
 
-
-.. _transforms_pil_only:
-
-Transforms on PIL Image only
-----------------------------
+Composition
+-----------
 
 .. autosummary::
     :toctree: generated/
     :template: class.rst
 
+    Compose
+    RandomApply
     RandomChoice
     RandomOrder
 
-.. _transforms_tensor_only:
-
-Transforms on torch.\*Tensor only
----------------------------------
+Miscellaneous
+-------------
 
 .. autosummary::
     :toctree: generated/
@@ -155,12 +150,12 @@ Transforms on torch.\*Tensor only
     LinearTransformation
     Normalize
     RandomErasing
-    ConvertImageDtype
+    Lambda
 
 .. _conversion_transforms:
 
-Conversion Transforms
----------------------
+Conversion
+----------
 
 .. autosummary::
     :toctree: generated/
@@ -169,20 +164,10 @@ Conversion Transforms
     ToPILImage
     ToTensor
     PILToTensor
+    ConvertImageDtype
 
-
-Generic Transforms
-------------------
-
-.. autosummary::
-    :toctree: generated/
-    :template: class.rst
-
-    Lambda
-
-
-Automatic Augmentation Transforms
----------------------------------
+Auto-Augmentation
+-----------------
 
 `AutoAugment <https://arxiv.org/pdf/1805.09501.pdf>`_ is a common Data Augmentation technique that can improve the accuracy of Image Classification models.
 Though the data augmentation policies are directly linked to their trained dataset, empirical studies show that