Remove num_channels argument

src/transformers/models/beit/image_processing_beit.py

@@ -232,13 +232,12 @@ def _preprocess_image(
         image_std: Optional[Union[float, List[float]]] = None,
         data_format: Optional[Union[str, ChannelDimension]] = None,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
     ) -> np.ndarray:
         """Preprocesses a single image."""
         # All transformations expect numpy arrays.
         image = to_numpy_array(image)
         if input_data_format is None:
-            input_data_format = infer_channel_dimension_format(image, num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(image)
         image = self._preprocess(
             image,
             do_reduce_labels=False,
@@ -322,7 +321,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: ChannelDimension = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -375,9 +373,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         size = size if size is not None else self.size
@@ -436,7 +431,6 @@ def preprocess(
                 image_std=image_std,
                 data_format=data_format,
                 input_data_format=input_data_format,
-                num_channels=num_channels,
             )
             for img in images
         ]

src/transformers/models/bit/image_processing_bit.py

@@ -177,7 +177,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -229,9 +228,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         size = size if size is not None else self.size
@@ -276,7 +272,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         if do_resize:
             images = [

src/transformers/models/blip/image_processing_blip.py

@@ -169,7 +169,6 @@ def preprocess(
         do_convert_rgb: bool = None,
         data_format: ChannelDimension = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -217,9 +216,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         resample = resample if resample is not None else self.resample
@@ -259,7 +255,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         if do_resize:
             images = [

src/transformers/models/bridgetower/image_processing_bridgetower.py

@@ -380,7 +380,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: ChannelDimension = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -434,9 +433,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         size_divisor = size_divisor if size_divisor is not None else self.size_divisor

src/transformers/models/chinese_clip/image_processing_chinese_clip.py

@@ -175,7 +175,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -227,9 +226,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         size = size if size is not None else self.size
@@ -274,7 +270,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         if do_resize:
             images = [

src/transformers/models/clip/image_processing_clip.py

@@ -176,7 +176,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -228,9 +227,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         size = size if size is not None else self.size
@@ -275,7 +271,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         if do_resize:
             images = [

src/transformers/models/conditional_detr/image_processing_conditional_detr.py

@@ -1119,7 +1119,6 @@ def preprocess(
         return_tensors: Optional[Union[TensorType, str]] = None,
         data_format: Union[str, ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> BatchFeature:
         """
@@ -1176,9 +1175,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         if "pad_and_return_pixel_mask" in kwargs:
             logger.warning_once(
@@ -1265,7 +1261,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         # prepare (COCO annotations as a list of Dict -> DETR target as a single Dict per image)
         if annotations is not None:

src/transformers/models/convnext/image_processing_convnext.py

@@ -196,7 +196,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: ChannelDimension = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -245,9 +244,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         crop_pct = crop_pct if crop_pct is not None else self.crop_pct
@@ -286,7 +282,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         if do_resize:
             images = [

src/transformers/models/deformable_detr/image_processing_deformable_detr.py

@@ -1117,7 +1117,6 @@ def preprocess(
         return_tensors: Optional[Union[TensorType, str]] = None,
         data_format: Union[str, ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> BatchFeature:
         """
@@ -1174,9 +1173,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         if "pad_and_return_pixel_mask" in kwargs:
             logger.warning_once(
@@ -1263,7 +1259,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         # prepare (COCO annotations as a list of Dict -> DETR target as a single Dict per image)
         if annotations is not None:

src/transformers/models/deit/image_processing_deit.py

@@ -173,7 +173,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: ChannelDimension = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -221,9 +220,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         resample = resample if resample is not None else self.resample
@@ -264,7 +260,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         if do_resize:
             images = [

src/transformers/models/deta/image_processing_deta.py

@@ -789,7 +789,6 @@ def preprocess(
         return_tensors: Optional[Union[TensorType, str]] = None,
         data_format: Union[str, ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> BatchFeature:
         """
@@ -846,9 +845,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         if "pad_and_return_pixel_mask" in kwargs:
             logger.warning_once(
@@ -927,7 +923,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         # prepare (COCO annotations as a list of Dict -> DETR target as a single Dict per image)
         if annotations is not None:

src/transformers/models/detr/image_processing_detr.py

@@ -1089,7 +1089,6 @@ def preprocess(
         return_tensors: Optional[Union[TensorType, str]] = None,
         data_format: Union[str, ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> BatchFeature:
         """
@@ -1146,9 +1145,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset.
         """
         if "pad_and_return_pixel_mask" in kwargs:
             logger.warning_once(
@@ -1235,7 +1231,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         # prepare (COCO annotations as a list of Dict -> DETR target as a single Dict per image)
         if annotations is not None:

src/transformers/models/donut/image_processing_donut.py

@@ -312,7 +312,6 @@ def preprocess(
         return_tensors: Optional[Union[str, TensorType]] = None,
         data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
         input_data_format: Optional[Union[str, ChannelDimension]] = None,
-        num_channels: Optional[int] = None,
         **kwargs,
     ) -> PIL.Image.Image:
         """
@@ -368,9 +367,6 @@ def preprocess(
                 - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
                 - `"channels_last"` or `ChannelDimension.LAST`: image in (height, width, num_channels) format.
                 - `"none"` or `ChannelDimension.NONE`: image in (height, width) format.
-            num_channels (`int`, *optional*):
-                The number of channels in the input image, used to infer the channel dimension format if
-                `input_data_format` is unset. if `input_data_format` is unset.
         """
         do_resize = do_resize if do_resize is not None else self.do_resize
         size = size if size is not None else self.size
@@ -413,7 +409,7 @@ def preprocess(
 
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
-            input_data_format = infer_channel_dimension_format(images[0], num_channels=num_channels)
+            input_data_format = infer_channel_dimension_format(images[0])
 
         if do_align_long_axis:
             images = [self.align_long_axis(image, size=size, input_data_format=input_data_format) for image in images]