Uniformize model processors (models *with* special arg names)

src/transformers/models/clipseg/processing_clipseg.py

@@ -16,10 +16,23 @@
 Image/Text processor class for CLIPSeg
 """
 
+import sys
 import warnings
+from typing import List, Optional, Union
 
-from ...processing_utils import ProcessorMixin
-from ...tokenization_utils_base import BatchEncoding
+from ...image_utils import ImageInput
+from ...processing_utils import ProcessingKwargs, ProcessorMixin
+from ...tokenization_utils_base import BatchEncoding, PreTokenizedInput, TextInput
+
+
+if sys.version_info >= (3, 11):
+    from typing import Unpack
+else:
+    from typing_extensions import Unpack
+
+
+class CLIPSegProcessorKwargs(ProcessingKwargs, total=False):
+    _defaults = {}
 
 
 class CLIPSegProcessor(ProcessorMixin):
@@ -39,6 +52,8 @@ class CLIPSegProcessor(ProcessorMixin):
     attributes = ["image_processor", "tokenizer"]
     image_processor_class = "ViTImageProcessor"
     tokenizer_class = ("CLIPTokenizer", "CLIPTokenizerFast")
+    # For backward compatibility. See transformers.processing_utils.ProcessorMixin.prepare_and_validate_optional_call_args for more details.
+    optional_call_args = ["visual_prompt"]
 
     def __init__(self, image_processor=None, tokenizer=None, **kwargs):
         feature_extractor = None
@@ -58,7 +73,18 @@ def __init__(self, image_processor=None, tokenizer=None, **kwargs):
 
         super().__init__(image_processor, tokenizer)
 
-    def __call__(self, text=None, images=None, visual_prompt=None, return_tensors=None, **kwargs):
+    def __call__(
+        self,
+        text: Optional[Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]]] = None,
+        images: Optional[ImageInput] = None,
+        # The following is to capture `visual_prompt` argument that may be passed as a positional argument.
+        # See transformers.processing_utils.ProcessorMixin.prepare_and_validate_optional_call_args for more details.
+        # This behavior is only needed for backward compatibility and will be removed in future versions.
+        *args,
+        audio=None,
+        videos=None,
+        **kwargs: Unpack[CLIPSegProcessorKwargs],
+    ):
         """
         Main method to prepare for the model one or several sequences(s) and image(s). This method forwards the `text`
         and `kwargs` arguments to CLIPTokenizerFast's [`~CLIPTokenizerFast.__call__`] if `text` is not `None` to encode
@@ -79,14 +105,6 @@ def __call__(self, text=None, images=None, visual_prompt=None, return_tensors=No
                 NumPy array or PyTorch tensor. In case of a NumPy array/PyTorch tensor, each image should be of shape
                 (C, H, W), where C is a number of channels, H and W are image height and width.
 
-            return_tensors (`str` or [`~utils.TensorType`], *optional*):
-                If set, will return tensors of a particular framework. Acceptable values are:
-
-                - `'tf'`: Return TensorFlow `tf.constant` objects.
-                - `'pt'`: Return PyTorch `torch.Tensor` objects.
-                - `'np'`: Return NumPy `np.ndarray` objects.
-                - `'jax'`: Return JAX `jnp.ndarray` objects.
-
         Returns:
             [`BatchEncoding`]: A [`BatchEncoding`] with the following fields:
 
@@ -96,20 +114,30 @@ def __call__(self, text=None, images=None, visual_prompt=None, return_tensors=No
               `None`).
             - **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`.
         """
+
+        output_kwargs = self._merge_kwargs(
+            CLIPSegProcessorKwargs,
+            tokenizer_init_kwargs=self.tokenizer.init_kwargs,
+            **kwargs,
+            **self.prepare_and_validate_optional_call_args(*args),
+        )
+
+        visual_prompt = output_kwargs["images_kwargs"].pop("visual_prompt", None)
+
         if text is None and visual_prompt is None and images is None:
             raise ValueError("You have to specify either text, visual prompt or images.")
 
         if text is not None and visual_prompt is not None:
             raise ValueError("You have to specify exactly one type of prompt. Either text or visual prompt.")
 
         if text is not None:
-            encoding = self.tokenizer(text, return_tensors=return_tensors, **kwargs)
+            encoding = self.tokenizer(text, **output_kwargs["text_kwargs"])
 
         if visual_prompt is not None:
-            prompt_features = self.image_processor(visual_prompt, return_tensors=return_tensors, **kwargs)
+            prompt_features = self.image_processor(visual_prompt, **output_kwargs["images_kwargs"])
 
         if images is not None:
-            image_features = self.image_processor(images, return_tensors=return_tensors, **kwargs)
+            image_features = self.image_processor(images, **output_kwargs["images_kwargs"])
 
         if visual_prompt is not None and images is not None:
             encoding = {
@@ -128,7 +156,9 @@ def __call__(self, text=None, images=None, visual_prompt=None, return_tensors=No
             }
             return encoding
         else:
-            return BatchEncoding(data=dict(**image_features), tensor_type=return_tensors)
+            return BatchEncoding(
+                data=dict(**image_features), tensor_type=output_kwargs["common_kwargs"].get("return_tensors")
+            )
 
     def batch_decode(self, *args, **kwargs):
         """

src/transformers/models/donut/image_processing_donut.py

@@ -183,6 +183,7 @@ def pad_image(
             input_data_format (`ChannelDimension` or `str`, *optional*):
                 The channel dimension format of the input image. If not provided, it will be inferred.
         """
+        size = get_size_dict(size)
         output_height, output_width = size["height"], size["width"]
         input_height, input_width = get_image_size(image, channel_dim=input_data_format)
 
@@ -232,6 +233,7 @@ def thumbnail(
                 The channel dimension format of the input image. If not provided, it will be inferred.
         """
         input_height, input_width = get_image_size(image, channel_dim=input_data_format)
+        size = get_size_dict(size)
         output_height, output_width = size["height"], size["width"]
 
         # We always resize to the smallest of either the input or output size.

src/transformers/models/fuyu/image_processing_fuyu.py

@@ -19,7 +19,7 @@
 
 import numpy as np
 
-from ...image_processing_utils import BaseImageProcessor, BatchFeature
+from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
 from ...image_transforms import (
     pad,
     resize,
@@ -344,6 +344,7 @@ def pad_image(
                 The channel dimension format of the input image. If not provided, it will be inferred.
         """
         image_height, image_width = get_image_size(image, input_data_format)
+        size = get_size_dict(size)
         target_height, target_width = size["height"], size["width"]
         padding_top = 0
         padding_left = 0

src/transformers/models/nougat/image_processing_nougat.py

@@ -250,6 +250,7 @@ def pad_image(
             input_data_format (`ChannelDimension` or `str`, *optional*):
                 The channel dimension format of the input image. If not provided, it will be inferred.
         """
+        size = get_size_dict(size)
         output_height, output_width = size["height"], size["width"]
         input_height, input_width = get_image_size(image, channel_dim=input_data_format)
 
@@ -292,6 +293,7 @@ def thumbnail(
                 The channel dimension format of the input image. If not provided, it will be inferred.
         """
         input_height, input_width = get_image_size(image, channel_dim=input_data_format)
+        size = get_size_dict(size)
         output_height, output_width = size["height"], size["width"]
 
         # We always resize to the smallest of either the input or output size.

src/transformers/models/nougat/processing_nougat.py

@@ -16,12 +16,44 @@
 Processor class for Nougat.
 """
 
-from typing import Dict, List, Optional, Union
+import sys
+from typing import List, Optional, Union
 
-from transformers.tokenization_utils_base import PreTokenizedInput, TextInput, TruncationStrategy
+from ...image_utils import ImageInput
+from ...processing_utils import ImagesKwargs, ProcessingKwargs, ProcessorMixin
+from ...tokenization_utils_base import PreTokenizedInput, TextInput
 
-from ...processing_utils import ProcessorMixin
-from ...utils import PaddingStrategy, TensorType
+
+if sys.version_info >= (3, 11):
+    from typing import Unpack
+else:
+    from typing_extensions import Unpack
+
+
+class NougatImagesKwargs(ImagesKwargs, total=False):
+    do_crop_margin: Optional[bool]
+    do_thumbnail: Optional[bool]
+    do_align_long_axis: Optional[bool]
+
+
+class NougatProcessorKwargs(ProcessingKwargs, total=False):
+    images_kwargs: NougatImagesKwargs
+    _defaults = {
+        "text_kwargs": {
+            "add_special_tokens": True,
+            "padding": False,
+            "stride": 0,
+            "is_split_into_words": False,
+            "return_overflowing_tokens": False,
+            "return_special_tokens_mask": False,
+            "return_offsets_mapping": False,
+            "return_length": False,
+            "verbose": True,
+        },
+        "images_kwargs": {
+            "data_format": "channels_first",
+        },
+    }
 
 
 class NougatProcessor(ProcessorMixin):
@@ -48,87 +80,27 @@ def __init__(self, image_processor, tokenizer):
 
     def __call__(
         self,
-        images=None,
-        text=None,
-        do_crop_margin: bool = None,
-        do_resize: bool = None,
-        size: Dict[str, int] = None,
-        resample: "PILImageResampling" = None,  # noqa: F821
-        do_thumbnail: bool = None,
-        do_align_long_axis: bool = None,
-        do_pad: bool = None,
-        do_rescale: bool = None,
-        rescale_factor: Union[int, float] = None,
-        do_normalize: bool = None,
-        image_mean: Optional[Union[float, List[float]]] = None,
-        image_std: Optional[Union[float, List[float]]] = None,
-        data_format: Optional["ChannelDimension"] = "channels_first",  # noqa: F821
-        input_data_format: Optional[Union[str, "ChannelDimension"]] = None,  # noqa: F821
-        text_pair: Optional[Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]]] = None,
-        text_target: Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]] = None,
-        text_pair_target: Optional[
-            Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]]
-        ] = None,
-        add_special_tokens: bool = True,
-        padding: Union[bool, str, PaddingStrategy] = False,
-        truncation: Union[bool, str, TruncationStrategy] = None,
-        max_length: Optional[int] = None,
-        stride: int = 0,
-        is_split_into_words: bool = False,
-        pad_to_multiple_of: Optional[int] = None,
-        return_tensors: Optional[Union[str, TensorType]] = None,
-        return_token_type_ids: Optional[bool] = None,
-        return_attention_mask: Optional[bool] = None,
-        return_overflowing_tokens: bool = False,
-        return_special_tokens_mask: bool = False,
-        return_offsets_mapping: bool = False,
-        return_length: bool = False,
-        verbose: bool = True,
+        images: Optional[ImageInput] = None,
+        text: Optional[Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]]] = None,
+        audio=None,
+        videos=None,
+        **kwargs: Unpack[NougatProcessorKwargs],
     ):
         if images is None and text is None:
             raise ValueError("You need to specify either an `images` or `text` input to process.")
 
+        output_kwargs = self._merge_kwargs(
+            NougatProcessorKwargs,
+            tokenizer_init_kwargs=self.tokenizer.init_kwargs,
+            **kwargs,
+        )
+        # Temporary fix for "paddding_side" in init_kwargs
+        _ = output_kwargs["text_kwargs"].pop("padding_side", None)
+
         if images is not None:
-            inputs = self.image_processor(
-                images,
-                do_crop_margin=do_crop_margin,
-                do_resize=do_resize,
-                size=size,
-                resample=resample,
-                do_thumbnail=do_thumbnail,
-                do_align_long_axis=do_align_long_axis,
-                do_pad=do_pad,
-                do_rescale=do_rescale,
-                rescale_factor=rescale_factor,
-                do_normalize=do_normalize,
-                image_mean=image_mean,
-                image_std=image_std,
-                return_tensors=return_tensors,
-                data_format=data_format,
-                input_data_format=input_data_format,
-            )
+            inputs = self.image_processor(images, **output_kwargs["images_kwargs"])
         if text is not None:
-            encodings = self.tokenizer(
-                text,
-                text_pair=text_pair,
-                text_target=text_target,
-                text_pair_target=text_pair_target,
-                add_special_tokens=add_special_tokens,
-                padding=padding,
-                truncation=truncation,
-                max_length=max_length,
-                stride=stride,
-                is_split_into_words=is_split_into_words,
-                pad_to_multiple_of=pad_to_multiple_of,
-                return_tensors=return_tensors,
-                return_token_type_ids=return_token_type_ids,
-                return_attention_mask=return_attention_mask,
-                return_overflowing_tokens=return_overflowing_tokens,
-                return_special_tokens_mask=return_special_tokens_mask,
-                return_offsets_mapping=return_offsets_mapping,
-                return_length=return_length,
-                verbose=verbose,
-            )
+            encodings = self.tokenizer(text, **output_kwargs["text_kwargs"])
 
         if text is None:
             return inputs

src/transformers/models/owlv2/image_processing_owlv2.py

@@ -19,7 +19,7 @@
 
 import numpy as np
 
-from ...image_processing_utils import BaseImageProcessor, BatchFeature
+from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
 from ...image_transforms import (
     center_to_corners_format,
     pad,
@@ -296,6 +296,7 @@ def resize(
         """
         requires_backends(self, "scipy")
 
+        size = get_size_dict(size)
         output_shape = (size["height"], size["width"])
         image = to_channel_dimension_format(image, ChannelDimension.LAST)
         image, output_shape = _preprocess_resize_output_shape(image, output_shape)