Add ImageProcessorFast to Qwen2.5-VL processor

docs/source/en/model_doc/qwen2_5_vl.md

@@ -285,6 +285,11 @@ model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
 [[autodoc]] Qwen2_5_VLImageProcessor
     - preprocess
 
+## Qwen2_5_VLImageProcessorFast
+
+[[autodoc]] Qwen2_5_VLImageProcessorFast
+    - preprocess
+
 ## Qwen2_5_VLProcessor
 
 [[autodoc]] Qwen2_5_VLProcessor

src/transformers/__init__.py

@@ -1322,6 +1322,7 @@
     _import_structure["models.llava_next"].append("LlavaNextImageProcessorFast")
     _import_structure["models.llava_onevision"].append("LlavaOnevisionImageProcessorFast")
     _import_structure["models.pixtral"].append("PixtralImageProcessorFast")
+    _import_structure["models.qwen2_5_vl"].append("Qwen2_5_VLImageProcessorFast")
     _import_structure["models.qwen2_vl"].append("Qwen2VLImageProcessorFast")
     _import_structure["models.rt_detr"].append("RTDetrImageProcessorFast")
     _import_structure["models.siglip"].append("SiglipImageProcessorFast")
@@ -6478,6 +6479,7 @@
         from .models.llava_next import LlavaNextImageProcessorFast
         from .models.llava_onevision import LlavaOnevisionImageProcessorFast
         from .models.pixtral import PixtralImageProcessorFast
+        from .models.qwen2_5_vl import Qwen2_5_VLImageProcessorFast
         from .models.qwen2_vl import Qwen2VLImageProcessorFast
         from .models.rt_detr import RTDetrImageProcessorFast
         from .models.siglip import SiglipImageProcessorFast

src/transformers/models/auto/image_processing_auto.py

@@ -127,6 +127,7 @@
             ("poolformer", ("PoolFormerImageProcessor",)),
             ("pvt", ("PvtImageProcessor",)),
             ("pvt_v2", ("PvtImageProcessor",)),
+            ("qwen2_5_vl", ("Qwen2_5_VLImageProcessorFast", "Qwen2_5_VLImageProcessorFast")),
             ("qwen2_vl", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
             ("regnet", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
             ("resnet", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),

src/transformers/models/qwen2_5_vl/__init__.py

@@ -20,6 +20,7 @@
 if TYPE_CHECKING:
     from .configuration_qwen2_5_vl import *
     from .image_processing_qwen2_5_vl import *
+    from .image_processing_qwen2_5_vl_fast import *
     from .modeling_qwen2_5_vl import *
     from .processing_qwen2_5_vl import *
 else:

src/transformers/models/qwen2_5_vl/configuration_qwen2_5_vl.py

@@ -23,8 +23,16 @@
 # 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.
+
 from ...configuration_utils import PretrainedConfig
 from ...modeling_rope_utils import rope_config_validation
+from ...utils import is_torchvision_available, is_torchvision_v2_available
+
+
+if is_torchvision_v2_available():
+    pass
+elif is_torchvision_available():
+    pass
 
 
 class Qwen2_5_VLVisionConfig(PretrainedConfig):

src/transformers/models/qwen2_5_vl/image_processing_qwen2_5_vl.py

@@ -48,7 +48,18 @@
     valid_images,
     validate_preprocess_arguments,
 )
-from ...utils import TensorType, logging
+from ...utils import (
+    TensorType,
+    is_torchvision_available,
+    is_torchvision_v2_available,
+    logging,
+)
+
+
+if is_torchvision_v2_available():
+    pass
+elif is_torchvision_available():
+    pass
 
 
 logger = logging.get_logger(__name__)