Add vision requirement to image transforms (#20712)

* Add require_vision decorator

* Fixup

* Use requires_backends

* Add requires_backend to utils functions

src/transformers/image_transforms.py

@@ -18,26 +18,28 @@ from typing import Iterable, List, Optional, Tuple, Union
 
 import numpy as np
 
-from transformers.utils import ExplicitEnum, TensorType
-from transformers.utils.import_utils import is_flax_available, is_tf_available, is_torch_available, is_vision_available
-
-
-if is_vision_available():
-    import PIL
-
-    from .image_utils import (
+from transformers.image_utils import (
     ChannelDimension,
-        PILImageResampling,
     get_channel_dimension_axis,
     get_image_size,
     infer_channel_dimension_format,
-        is_jax_tensor,
-        is_tf_tensor,
-        is_torch_tensor,
     to_numpy_array,
-    )
+)
+from transformers.utils import ExplicitEnum, TensorType, is_jax_tensor, is_tf_tensor, is_torch_tensor
+from transformers.utils.import_utils import (
+    is_flax_available,
+    is_tf_available,
+    is_torch_available,
+    is_vision_available,
+    requires_backends,
+)
 
 
+if is_vision_available():
+    import PIL
+
+    from .image_utils import PILImageResampling
+
 if is_torch_available():
     import torch
 
@@ -116,9 +118,9 @@ def rescale(
 
 
 def to_pil_image(
-    image: Union[np.ndarray, PIL.Image.Image, "torch.Tensor", "tf.Tensor", "jnp.ndarray"],
+    image: Union[np.ndarray, "PIL.Image.Image", "torch.Tensor", "tf.Tensor", "jnp.ndarray"],
     do_rescale: Optional[bool] = None,
-) -> PIL.Image.Image:
+) -> "PIL.Image.Image":
     """
     Converts `image` to a PIL Image. Optionally rescales it and puts the channel dimension back as the last axis if
     needed.
@@ -133,6 +135,8 @@ def to_pil_image(
     Returns:
         `PIL.Image.Image`: The converted image.
     """
+    requires_backends(to_pil_image, ["vision"])
+
     if isinstance(image, PIL.Image.Image):
         return image
 
@@ -226,7 +230,7 @@ def get_resize_output_image_size(
 def resize(
     image,
     size: Tuple[int, int],
-    resample=PILImageResampling.BILINEAR,
+    resample: "PILImageResampling" = None,
     reducing_gap: Optional[int] = None,
     data_format: Optional[ChannelDimension] = None,
     return_numpy: bool = True,
@@ -253,6 +257,10 @@ def resize(
     Returns:
         `np.ndarray`: The resized image.
     """
+    requires_backends(resize, ["vision"])
+
+    resample = resample if resample is not None else PILImageResampling.BILINEAR
+
     if not len(size) == 2:
         raise ValueError("size must have 2 elements")
 
@@ -303,6 +311,8 @@ def normalize(
         data_format (`ChannelDimension`, *optional*):
             The channel dimension format of the output image. If unset, will use the inferred format from the input.
     """
+    requires_backends(normalize, ["vision"])
+
     if isinstance(image, PIL.Image.Image):
         warnings.warn(
             "PIL.Image.Image inputs are deprecated and will be removed in v4.26.0. Please use numpy arrays instead.",
@@ -372,6 +382,8 @@ def center_crop(
     Returns:
         `np.ndarray`: The cropped image.
     """
+    requires_backends(center_crop, ["vision"])
+
     if isinstance(image, PIL.Image.Image):
         warnings.warn(
             "PIL.Image.Image inputs are deprecated and will be removed in v4.26.0. Please use numpy arrays instead.",

src/transformers/image_utils.py

@@ -28,6 +28,7 @@ from .utils import (
     is_torch_available,
     is_torch_tensor,
     is_vision_available,
+    requires_backends,
     to_numpy,
 )
 from .utils.constants import (  # noqa: F401
@@ -64,7 +65,8 @@ class ChannelDimension(ExplicitEnum):
 
 def is_valid_image(img):
     return (
-        isinstance(img, (PIL.Image.Image, np.ndarray))
+        (is_vision_available() and isinstance(img, PIL.Image.Image))
+        or isinstance(img, np.ndarray)
         or is_torch_tensor(img)
         or is_tf_tensor(img)
         or is_jax_tensor(img)
@@ -90,7 +92,10 @@ def is_batched(img):
 
 
 def to_numpy_array(img) -> np.ndarray:
-    if isinstance(img, PIL.Image.Image):
+    if not is_valid_image(img):
+        raise ValueError(f"Invalid image type: {type(img)}")
+
+    if is_vision_available() and isinstance(img, PIL.Image.Image):
         return np.array(img)
     return to_numpy(img)
 
@@ -215,6 +220,7 @@ def load_image(image: Union[str, "PIL.Image.Image"]) -> "PIL.Image.Image":
     Returns:
         `PIL.Image.Image`: A PIL Image.
     """
+    requires_backends(load_image, ["vision"])
     if isinstance(image, str):
         if image.startswith("http://") or image.startswith("https://"):
             # We need to actually check for a real protocol, otherwise it's impossible to use a local file