[fbsync] port FiveCrop and TenCrop to prototype API (#5513)

Summary:
* port FiveCrop and TenCrop to prototype API

* fix ten crop for pil

* Update torchvision/prototype/transforms/_geometry.py

* simplify implementation

* minor cleanup

Reviewed By: vmoens

Differential Revision: D34878994

fbshipit-source-id: d1220091f9da4515e831bc44e873041ab33f3ce0

Co-authored-by: Philip Meier <github.pmeier@posteo.de>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

Author: 2 people

torchvision/prototype/transforms/__init__.py

@@ -7,7 +7,7 @@
 from ._augment import RandomErasing, RandomMixup, RandomCutmix
 from ._auto_augment import RandAugment, TrivialAugmentWide, AutoAugment, AugMix
 from ._container import Compose, RandomApply, RandomChoice, RandomOrder
-from ._geometry import HorizontalFlip, Resize, CenterCrop, RandomResizedCrop
+from ._geometry import HorizontalFlip, Resize, CenterCrop, RandomResizedCrop, FiveCrop, TenCrop, BatchMultiCrop
 from ._meta import ConvertBoundingBoxFormat, ConvertImageDtype, ConvertImageColorSpace
 from ._misc import Identity, Normalize, ToDtype, Lambda
 from ._presets import CocoEval, ImageNetEval, VocEval, Kinect400Eval, RaftEval

torchvision/prototype/transforms/_geometry.py

@@ -1,3 +1,4 @@
+import collections.abc
 import math
 import warnings
 from typing import Any, Dict, List, Union, Sequence, Tuple, cast
@@ -6,6 +7,7 @@
 import torch
 from torchvision.prototype import features
 from torchvision.prototype.transforms import Transform, InterpolationMode, functional as F
+from torchvision.transforms.functional import pil_to_tensor
 from torchvision.transforms.transforms import _setup_size, _interpolation_modes_from_int
 
 from ._utils import query_image, get_image_dimensions, has_any, is_simple_tensor
@@ -168,3 +170,89 @@ def forward(self, *inputs: Any) -> Any:
         if has_any(sample, features.BoundingBox, features.SegmentationMask):
             raise TypeError(f"BoundingBox'es and SegmentationMask's are not supported by {type(self).__name__}()")
         return super().forward(sample)
+
+
+class MultiCropResult(list):
+    """Helper class for :class:`~torchvision.prototype.transforms.BatchMultiCrop`.
+
+    Outputs of multi crop transforms such as :class:`~torchvision.prototype.transforms.FiveCrop` and
+    `:class:`~torchvision.prototype.transforms.TenCrop` should be wrapped in this in order to be batched correctly by
+    :class:`~torchvision.prototype.transforms.BatchMultiCrop`.
+    """
+
+    pass
+
+
+class FiveCrop(Transform):
+    def __init__(self, size: Union[int, Sequence[int]]) -> None:
+        super().__init__()
+        self.size = _setup_size(size, error_msg="Please provide only two dimensions (h, w) for size.")
+
+    def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
+        if isinstance(input, features.Image):
+            output = F.five_crop_image_tensor(input, self.size)
+            return MultiCropResult(features.Image.new_like(input, o) for o in output)
+        elif is_simple_tensor(input):
+            return MultiCropResult(F.five_crop_image_tensor(input, self.size))
+        elif isinstance(input, PIL.Image.Image):
+            return MultiCropResult(F.five_crop_image_pil(input, self.size))
+        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__}()")
+        return super().forward(sample)
+
+
+class TenCrop(Transform):
+    def __init__(self, size: Union[int, Sequence[int]], vertical_flip: bool = False) -> None:
+        super().__init__()
+        self.size = _setup_size(size, error_msg="Please provide only two dimensions (h, w) for size.")
+        self.vertical_flip = vertical_flip
+
+    def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
+        if isinstance(input, features.Image):
+            output = F.ten_crop_image_tensor(input, self.size, vertical_flip=self.vertical_flip)
+            return MultiCropResult(features.Image.new_like(input, o) for o in output)
+        elif is_simple_tensor(input):
+            return MultiCropResult(F.ten_crop_image_tensor(input, self.size))
+        elif isinstance(input, PIL.Image.Image):
+            return MultiCropResult(F.ten_crop_image_pil(input, self.size))
+        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__}()")
+        return super().forward(sample)
+
+
+class BatchMultiCrop(Transform):
+    def forward(self, *inputs: Any) -> Any:
+        # This is basically the functionality of `torchvision.prototype.utils._internal.apply_recursively` with one
+        # significant difference:
+        # Since we need multiple images to batch them together, we need to explicitly exclude `MultiCropResult` from
+        # the sequence case.
+        def apply_recursively(obj: Any) -> Any:
+            if isinstance(obj, MultiCropResult):
+                crops = obj
+                if isinstance(obj[0], PIL.Image.Image):
+                    crops = [pil_to_tensor(crop) for crop in crops]  # type: ignore[assignment]
+
+                batch = torch.stack(crops)
+
+                if isinstance(obj[0], features.Image):
+                    batch = features.Image.new_like(obj[0], batch)
+
+                return batch
+            elif isinstance(obj, collections.abc.Sequence) and not isinstance(obj, str):
+                return [apply_recursively(item) for item in obj]
+            elif isinstance(obj, collections.abc.Mapping):
+                return {key: apply_recursively(item) for key, item in obj.items()}
+            else:
+                return obj
+
+        return apply_recursively(inputs if len(inputs) > 1 else inputs[0])

torchvision/prototype/transforms/functional/__init__.py

@@ -60,6 +60,10 @@
     perspective_image_pil,
     vertical_flip_image_tensor,
     vertical_flip_image_pil,
+    five_crop_image_tensor,
+    five_crop_image_pil,
+    ten_crop_image_tensor,
+    ten_crop_image_pil,
 )
 from ._misc import normalize_image_tensor, gaussian_blur_image_tensor
 from ._type_conversion import decode_image_with_pil, decode_video_with_av, label_to_one_hot

torchvision/prototype/transforms/functional/_geometry.py

@@ -314,3 +314,79 @@ def resized_crop_image_pil(
 ) -> PIL.Image.Image:
     img = crop_image_pil(img, top, left, height, width)
     return resize_image_pil(img, size, interpolation=interpolation)
+
+
+def _parse_five_crop_size(size: List[int]) -> List[int]:
+    if isinstance(size, numbers.Number):
+        size = (int(size), int(size))
+    elif isinstance(size, (tuple, list)) and len(size) == 1:
+        size = (size[0], size[0])  # type: ignore[assignment]
+
+    if len(size) != 2:
+        raise ValueError("Please provide only two dimensions (h, w) for size.")
+
+    return size
+
+
+def five_crop_image_tensor(
+    img: torch.Tensor, size: List[int]
+) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
+    crop_height, crop_width = _parse_five_crop_size(size)
+    _, image_height, image_width = get_dimensions_image_tensor(img)
+
+    if crop_width > image_width or crop_height > image_height:
+        msg = "Requested crop size {} is bigger than input size {}"
+        raise ValueError(msg.format(size, (image_height, image_width)))
+
+    tl = crop_image_tensor(img, 0, 0, crop_height, crop_width)
+    tr = crop_image_tensor(img, 0, image_width - crop_width, crop_height, crop_width)
+    bl = crop_image_tensor(img, image_height - crop_height, 0, crop_height, crop_width)
+    br = crop_image_tensor(img, image_height - crop_height, image_width - crop_width, crop_height, crop_width)
+    center = center_crop_image_tensor(img, [crop_height, crop_width])
+
+    return tl, tr, bl, br, center
+
+
+def five_crop_image_pil(
+    img: PIL.Image.Image, size: List[int]
+) -> Tuple[PIL.Image.Image, PIL.Image.Image, PIL.Image.Image, PIL.Image.Image, PIL.Image.Image]:
+    crop_height, crop_width = _parse_five_crop_size(size)
+    _, image_height, image_width = get_dimensions_image_pil(img)
+
+    if crop_width > image_width or crop_height > image_height:
+        msg = "Requested crop size {} is bigger than input size {}"
+        raise ValueError(msg.format(size, (image_height, image_width)))
+
+    tl = crop_image_pil(img, 0, 0, crop_height, crop_width)
+    tr = crop_image_pil(img, 0, image_width - crop_width, crop_height, crop_width)
+    bl = crop_image_pil(img, image_height - crop_height, 0, crop_height, crop_width)
+    br = crop_image_pil(img, image_height - crop_height, image_width - crop_width, crop_height, crop_width)
+    center = center_crop_image_pil(img, [crop_height, crop_width])
+
+    return tl, tr, bl, br, center
+
+
+def ten_crop_image_tensor(img: torch.Tensor, size: List[int], vertical_flip: bool = False) -> List[torch.Tensor]:
+    tl, tr, bl, br, center = five_crop_image_tensor(img, size)
+
+    if vertical_flip:
+        img = vertical_flip_image_tensor(img)
+    else:
+        img = horizontal_flip_image_tensor(img)
+
+    tl_flip, tr_flip, bl_flip, br_flip, center_flip = five_crop_image_tensor(img, size)
+
+    return [tl, tr, bl, br, center, tl_flip, tr_flip, bl_flip, br_flip, center_flip]
+
+
+def ten_crop_image_pil(img: PIL.Image.Image, size: List[int], vertical_flip: bool = False) -> List[PIL.Image.Image]:
+    tl, tr, bl, br, center = five_crop_image_pil(img, size)
+
+    if vertical_flip:
+        img = vertical_flip_image_pil(img)
+    else:
+        img = horizontal_flip_image_pil(img)
+
+    tl_flip, tr_flip, bl_flip, br_flip, center_flip = five_crop_image_pil(img, size)
+
+    return [tl, tr, bl, br, center, tl_flip, tr_flip, bl_flip, br_flip, center_flip]