[OpenVINO][Draft]add support for dinov3

optimum/exporters/openvino/model_configs.py

@@ -4537,3 +4537,25 @@ def patch_model_for_export(
         self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
     ) -> "ModelPatcher":
         return OVDecoderModelPatcher(self, model, model_kwargs=model_kwargs)
+
+
+@register_in_tasks_manager("dinov3_vit", *["feature-extraction"], library_name="transformers")
+@register_in_tasks_manager("dinov3_convnext", *["feature-extraction"], library_name="transformers")
+class OpenDinoV3OpenVINOConfig(VisionOnnxConfig):
+    MIN_TRANSFORMERS_VERSION = version.parse("4.55.0")
+
+    DEFAULT_ONNX_OPSET = 14
+
+    NORMALIZED_CONFIG_CLASS = NormalizedVisionConfig
+    @property
+    def inputs(self) -> Dict[str, Dict[int, str]]:
+        return {
+            "pixel_values": {0: "image_batch_size", 1: "num_channels", 2: "height", 3: "width"},
+        }
+
+    @property
+    def outputs(self) -> Dict[str, Dict[int, str]]:
+        return {
+            "last_hidden_state": {0: "image_batch_size"},
+            "pooler_output": {0: "image_batch_size"},
+        }

tests/openvino/test_exporters_cli.py

@@ -121,6 +121,13 @@ class OVCLIExportTestCase(unittest.TestCase):
                 ("text-generation-with-past", "ernie4_5"),
             ]
         )
+
+    if is_transformers_version(">=", "4.56"):
+        SUPPORTED_ARCHITECTURES.extend(
+            [
+                ("feature-extraction", "vit-with-pooler"),
+            ]
+        )
 
     EXPECTED_NUMBER_OF_TOKENIZER_MODELS = {
         "gpt2": 2 if is_tokenizers_version("<", "0.20") or is_openvino_version(">=", "2024.5") else 0,

tests/openvino/utils_tests.py

@@ -197,6 +197,7 @@
     "sana": "katuni4ka/tiny-random-sana",
     "sana-sprint": "katuni4ka/tiny-random-sana-sprint",
     "ltx-video": "katuni4ka/tiny-random-ltx-video",
+    "vit-with-pooler": "snake7gun/tiny-random-dinov3",
 }