[Fast Processor] OWLv2

docs/source/en/model_doc/owlv2.md

@@ -106,6 +106,11 @@ Usage of OWLv2 is identical to [OWL-ViT](owlvit) with a new, updated image proce
     - post_process_object_detection
     - post_process_image_guided_detection
 
+## Owlv2ImageProcessorFast
+
+[[autodoc]] Owlv2ImageProcessorFast
+    - preprocess
+
 ## Owlv2Processor
 
 [[autodoc]] Owlv2Processor

src/transformers/__init__.py

@@ -1357,6 +1357,7 @@
     _import_structure["models.llava"].append("LlavaImageProcessorFast")
     _import_structure["models.llava_next"].append("LlavaNextImageProcessorFast")
     _import_structure["models.llava_onevision"].append("LlavaOnevisionImageProcessorFast")
+    _import_structure["models.owlv2"].append("Owlv2ImageProcessorFast")
     _import_structure["models.phi4_multimodal"].append("Phi4MultimodalImageProcessorFast")
     _import_structure["models.pixtral"].append("PixtralImageProcessorFast")
     _import_structure["models.qwen2_vl"].append("Qwen2VLImageProcessorFast")
@@ -6649,6 +6650,7 @@
         from .models.llava import LlavaImageProcessorFast
         from .models.llava_next import LlavaNextImageProcessorFast
         from .models.llava_onevision import LlavaOnevisionImageProcessorFast
+        from .models.owlv2 import Owlv2ImageProcessorFast
         from .models.phi4_multimodal import Phi4MultimodalImageProcessorFast
         from .models.pixtral import PixtralImageProcessorFast
         from .models.qwen2_vl import Qwen2VLImageProcessorFast

src/transformers/models/auto/image_processing_auto.py

@@ -120,7 +120,7 @@
             ("nat", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("nougat", ("NougatImageProcessor",)),
             ("oneformer", ("OneFormerImageProcessor",)),
-            ("owlv2", ("Owlv2ImageProcessor",)),
+            ("owlv2", ("Owlv2ImageProcessor", "Owlv2ImageProcessorFast")),
             ("owlvit", ("OwlViTImageProcessor",)),
             ("paligemma", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
             ("perceiver", ("PerceiverImageProcessor",)),

src/transformers/models/owlv2/__init__.py

@@ -20,6 +20,7 @@
 if TYPE_CHECKING:
     from .configuration_owlv2 import *
     from .image_processing_owlv2 import *
+    from .image_processing_owlv2_fast import *
     from .modeling_owlv2 import *
     from .processing_owlv2 import *
 else:

src/transformers/models/owlv2/image_processing_owlv2_fast.py

@@ -0,0 +1,58 @@
+# coding=utf-8
+# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Fast Image processor class for OWLv2."""
+
+from ...image_processing_utils_fast import (
+    BASE_IMAGE_PROCESSOR_FAST_DOCSTRING,
+    BaseImageProcessorFast,
+    DefaultFastImageProcessorKwargs,
+    Unpack,
+)
+from ...image_utils import OPENAI_CLIP_MEAN, OPENAI_CLIP_STD, PILImageResampling
+from ...utils import add_start_docstrings
+
+
+@add_start_docstrings(
+    "Constructs a fast Owlv2 image processor.",
+    BASE_IMAGE_PROCESSOR_FAST_DOCSTRING,
+)
+class Owlv2ImageProcessorFast(BaseImageProcessorFast):
+    # This generated class can be used as a starting point for the fast image processor.
+    # if the image processor is only used for simple augmentations, such as resizing, center cropping, rescaling, or normalizing,
+    # only the default values should be set in the class.
+    # If the image processor requires more complex augmentations, methods from BaseImageProcessorFast can be overridden.
+    # In most cases, only the `_preprocess` method should be overridden.
+
+    # For an example of a fast image processor requiring more complex augmentations, see `LlavaNextImageProcessorFast`.
+
+    # Default values should be checked against the slow image processor
+    # None values left after checking can be removed
+    resample = PILImageResampling.BILINEAR
+    image_mean = OPENAI_CLIP_MEAN
+    image_std = OPENAI_CLIP_STD
+    size = {"height": 960, "width": 960}
+    default_to_square = None
+    crop_size = None
+    do_resize = True
+    do_center_crop = None
+    do_rescale = True
+    do_normalize = True
+    do_convert_rgb = None
+
+    def __init__(self, **kwargs: Unpack[DefaultFastImageProcessorKwargs]) -> None:
+        super().__init__(**kwargs)
+
+
+__all__ = ["Owlv2ImageProcessorFast"]

src/transformers/utils/dummy_torchvision_objects.py

@@ -93,6 +93,13 @@ def __init__(self, *args, **kwargs):
         requires_backends(self, ["torchvision"])
 
 
+class Owlv2ImageProcessorFast(metaclass=DummyObject):
+    _backends = ["torchvision"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torchvision"])
+
+
 class Phi4MultimodalImageProcessorFast(metaclass=DummyObject):
     _backends = ["torchvision"]
 

tests/models/owlv2/test_image_processing_owlv2.py

@@ -17,15 +17,15 @@
 import unittest
 
 from transformers.testing_utils import require_torch, require_vision, slow
-from transformers.utils import is_torch_available, is_vision_available
+from transformers.utils import is_torch_available, is_torchvision_available, is_vision_available
 
 from ...test_image_processing_common import ImageProcessingTestMixin, prepare_image_inputs
 
 
 if is_vision_available():
     from PIL import Image
 
-    from transformers import AutoProcessor, Owlv2ForObjectDetection, Owlv2ImageProcessor
+    from transformers import AutoProcessor, Owlv2ForObjectDetection, Owlv2ImageProcessor, Owlv2ImageProcessorFast
 
 if is_torch_available():
     import torch
@@ -88,6 +88,7 @@ def prepare_image_inputs(self, equal_resolution=False, numpify=False, torchify=F
 @require_vision
 class Owlv2ImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
     image_processing_class = Owlv2ImageProcessor if is_vision_available() else None
+    fast_image_processing_class = Owlv2ImageProcessorFast if is_torchvision_available() else None
 
     def setUp(self):
         super().setUp()