pylint & update model api version in docstring

src/otx/algorithms/visual_prompting/adapters/openvino/__init__.py

@@ -14,4 +14,4 @@
 # See the License for the specific language governing permissions
 # and limitations under the License.
 
-from .model_wrappers import *
+from .model_wrappers import *  # noqa: F403

src/otx/algorithms/visual_prompting/adapters/openvino/model_wrappers/__init__.py

@@ -14,5 +14,5 @@
 # See the License for the specific language governing permissions
 # and limitations under the License.
 
-from .openvino_models import ImageEncoder, Decoder  # noqa: F401
 from .openvino_adapters import VisualPromptingOpenvinoAdapter  # noqa: F401
+from .openvino_models import Decoder, ImageEncoder  # noqa: F401

src/otx/algorithms/visual_prompting/adapters/openvino/model_wrappers/openvino_adapters.py

@@ -25,8 +25,7 @@
 
 
 def resize_image_with_aspect_pad(input: Output, size, keep_aspect_ratio, interpolation, pad_value):
-    """https://github.com/openvinotoolkit/model_api/blob/master/model_api/python/openvino/model_api/adapters/utils.py#L273-L341
-    """
+    """https://github.com/openvinotoolkit/model_api/blob/0.1.3/model_api/python/openvino/model_api/adapters/utils.py#L273-L341."""
     h_axis = 1
     w_axis = 2
     w, h = size
@@ -46,12 +45,8 @@ def resize_image_with_aspect_pad(input: Output, size, keep_aspect_ratio, interpo
     w_ratio = opset.divide(np.float32(w), iw)
     h_ratio = opset.divide(np.float32(h), ih)
     scale = opset.minimum(w_ratio, h_ratio)
-    nw = opset.convert(
-        opset.round(opset.multiply(iw, scale), "half_to_even"), destination_type="i32"
-    )
-    nh = opset.convert(
-        opset.round(opset.multiply(ih, scale), "half_to_even"), destination_type="i32"
-    )
+    nw = opset.convert(opset.round(opset.multiply(iw, scale), "half_to_even"), destination_type="i32")
+    nh = opset.convert(opset.round(opset.multiply(ih, scale), "half_to_even"), destination_type="i32")
     new_size = opset.concat([opset.unsqueeze(nh, 0), opset.unsqueeze(nw, 0)], axis=0)
     image = opset.interpolate(
         input,
@@ -84,8 +79,7 @@ def resize_image_with_aspect_pad(input: Output, size, keep_aspect_ratio, interpo
 
 
 def resize_image_with_aspect(size, interpolation, pad_value):
-    """https://github.com/openvinotoolkit/model_api/blob/master/model_api/python/openvino/model_api/adapters/utils.py#L356-L365
-    """
+    """https://github.com/openvinotoolkit/model_api/blob/0.1.3/model_api/python/openvino/model_api/adapters/utils.py#L356-L365."""
     return custom_preprocess_function(
         partial(
             resize_image_with_aspect_pad,
@@ -99,10 +93,11 @@ def resize_image_with_aspect(size, interpolation, pad_value):
 
 class VisualPromptingOpenvinoAdapter(OpenvinoAdapter):
     """Openvino Adapter Wrappers of OTX Visual Prompting.
-    
+
     This class is to use fixed `fit_to_window` resize module.
     When model API version in otx is upgraded, it can be removed.
     """
+
     def embed_preprocessing(
         self,
         layout,
@@ -116,17 +111,14 @@ def embed_preprocessing(
         scale=None,
         input_idx=0,
     ):
-        """https://github.com/openvinotoolkit/model_api/blob/master/model_api/python/openvino/model_api/adapters/openvino_adapter.py#L340-L411
-        """
-        ppp = PrePostProcessor(self.model)
+        """https://github.com/openvinotoolkit/model_api/blob/0.1.3/model_api/python/openvino/model_api/adapters/openvino_adapter.py#L340-L411."""
+        ppp = PrePostProcessor(self.model)  # type: ignore[has-type]
 
         # Change the input type to the 8-bit image
         if dtype == type(int):
             ppp.input(input_idx).tensor().set_element_type(Type.u8)
 
-        ppp.input(input_idx).tensor().set_layout(ov.Layout("NHWC")).set_color_format(
-            ColorFormat.BGR
-        )
+        ppp.input(input_idx).tensor().set_layout(ov.Layout("NHWC")).set_color_format(ColorFormat.BGR)
 
         INTERPOLATION_MODE_MAP = {
             "LINEAR": "linear",
@@ -152,9 +144,7 @@ def embed_preprocessing(
                 )
 
             else:
-                raise ValueError(
-                    f"Upsupported resize type in model preprocessing: {resize_mode}"
-                )
+                raise ValueError(f"Upsupported resize type in model preprocessing: {resize_mode}")
 
         # Handle layout
         ppp.input(input_idx).model().set_layout(ov.Layout(layout))

src/otx/algorithms/visual_prompting/adapters/openvino/model_wrappers/openvino_models.py

@@ -20,7 +20,7 @@
 import cv2
 import numpy as np
 from openvino.model_api.adapters.inference_adapter import InferenceAdapter
-from openvino.model_api.models import DetectionModel, ImageModel
+from openvino.model_api.models import ImageModel
 from openvino.model_api.models.types import (
     BooleanValue,
     ListValue,
@@ -96,18 +96,20 @@ def preprocess(self, inputs: Dict[str, Any], meta: Dict[str, Any]):
             # TODO (sungchul): add condition to check whether using bbox or point
             point_coords = self._apply_coords(bbox.reshape(-1, 2, 2), inputs["original_size"])
             point_labels = np.array([2, 3], dtype=np.float32).reshape((-1, 2))
-            processed_prompts.append({
-                "point_coords": point_coords,
-                "point_labels": point_labels,
-                # TODO (sungchul): how to generate mask_input and has_mask_input
-                "mask_input": np.zeros((1, 1, 256, 256), dtype=np.float32),
-                "has_mask_input": np.zeros((1, 1), dtype=np.float32),
-                "orig_size": np.array(inputs["original_size"], dtype=np.float32).reshape((-1, 2)),
-                "label": label
-            })
+            processed_prompts.append(
+                {
+                    "point_coords": point_coords,
+                    "point_labels": point_labels,
+                    # TODO (sungchul): how to generate mask_input and has_mask_input
+                    "mask_input": np.zeros((1, 1, 256, 256), dtype=np.float32),
+                    "has_mask_input": np.zeros((1, 1), dtype=np.float32),
+                    "orig_size": np.array(inputs["original_size"], dtype=np.float32).reshape((-1, 2)),
+                    "label": label,
+                }
+            )
         return processed_prompts
 
-    def _apply_coords(self, coords: np.ndarray, original_size: Union[List[int], Tuple[int]]) -> np.ndarray:
+    def _apply_coords(self, coords: np.ndarray, original_size: Union[List[int], Tuple[int, int]]) -> np.ndarray:
         """Process coords according to preprocessed image size using image meta."""
         old_h, old_w = original_size
         new_h, new_w = self._get_preprocess_shape(original_size[0], original_size[1], self.image_size)
@@ -123,7 +125,7 @@ def _get_preprocess_shape(self, old_h: int, old_w: int, image_size: int) -> Tupl
         new_w = int(new_w + 0.5)
         new_h = int(new_h + 0.5)
         return (new_h, new_w)
-    
+
     def _check_io_number(self, number_of_inputs, number_of_outputs):
         pass
 
@@ -146,7 +148,7 @@ def postprocess(self, outputs: Dict[str, np.ndarray], meta: Dict[str, Any]) -> T
         """
 
         def sigmoid(x):
-            return np.tanh(x * 0.5) * 0.5 + 0.5 # to avoid overflow
+            return np.tanh(x * 0.5) * 0.5 + 0.5  # to avoid overflow
 
         soft_prediction = outputs[self.output_blob_name].squeeze()
         soft_prediction = self.resize_and_crop(soft_prediction, meta["original_size"][0])

src/otx/algorithms/visual_prompting/configs/base/configuration.py

@@ -19,11 +19,12 @@
 
 from otx.algorithms.common.configs import BaseConfig
 from otx.api.configuration.elements import (
+    ParameterGroup,
     add_parameter_group,
+    configurable_boolean,
     configurable_float,
     configurable_integer,
     string_attribute,
-    configurable_boolean
 )
 from otx.api.configuration.model_lifecycle import ModelLifecycle
 
@@ -41,7 +42,7 @@ class __LearningParameters(BaseConfig.BaseLearningParameters):
         description = header
 
     @attrs
-    class __Postprocessing:
+    class __Postprocessing(ParameterGroup):
         header = string_attribute("Postprocessing")
         description = header
 
@@ -86,7 +87,7 @@ class __Postprocessing:
             default_value=64,
             affects_outcome_of=ModelLifecycle.INFERENCE,
         )
-        
+
         orig_height = configurable_integer(
             header="Original height",
             description="Model input height before embedding processing.",

src/otx/algorithms/visual_prompting/tasks/openvino.py

@@ -24,11 +24,12 @@
 
 import attr
 import numpy as np
-from openvino.model_api.adapters import OpenvinoAdapter, create_core
+from openvino.model_api.adapters import create_core
 from openvino.model_api.models import Model
 
 from otx.algorithms.common.utils.logger import get_logger
 from otx.algorithms.common.utils.utils import get_default_async_reqs_num
+from otx.algorithms.visual_prompting.adapters.openvino import model_wrappers
 from otx.algorithms.visual_prompting.adapters.openvino.model_wrappers import (
     VisualPromptingOpenvinoAdapter,
 )

tests/unit/algorithms/visual_prompting/adapters/openvino/model_wrappers/test_openvino_models.py

@@ -101,19 +101,20 @@ def test_apply_coords(self):
         results = self.decoder._apply_coords(coords, original_size)
 
         assert results.shape == (1, 2, 2)
-        assert np.all(results == np.array([[[0.5, 0.5], [1., 1.]]]))
+        assert np.all(results == np.array([[[0.5, 0.5], [1.0, 1.0]]]))
 
     @e2e_pytest_unit
-    @pytest.mark.parametrize("old_h,old_w,image_size,expected",
+    @pytest.mark.parametrize(
+        "old_h,old_w,image_size,expected",
         [
             (4, 3, 6, (6, 5)),
             (3, 4, 6, (5, 6)),
-        ]
+        ],
     )
     def test_get_preprocess_shape(self, old_h: int, old_w: int, image_size: int, expected: Tuple[int]):
         """Test _get_preprocess_shape."""
         result = self.decoder._get_preprocess_shape(old_h, old_w, image_size)
-        
+
         assert result == expected
 
     @e2e_pytest_unit

tests/unit/algorithms/visual_prompting/adapters/pytorch_lightning/models/visual_prompters/test_segment_anything.py

@@ -286,7 +286,9 @@ def test_load_checkpoint_from_local_checkpoint(self, mocker, monkeypatch, checkp
         )
         mocker.patch("builtins.open").__enter__.return_value = True
         mocker.patch("torch.load", return_value=OrderedDict())
-        mocker_load_from_checkpoint = mocker.patch("otx.algorithms.visual_prompting.adapters.pytorch_lightning.models.visual_prompters.segment_anything.SegmentAnything.load_from_checkpoint")
+        mocker_load_from_checkpoint = mocker.patch(
+            "otx.algorithms.visual_prompting.adapters.pytorch_lightning.models.visual_prompters.segment_anything.SegmentAnything.load_from_checkpoint"
+        )
         mocker_load_state_dict = mocker.patch(
             "otx.algorithms.visual_prompting.adapters.pytorch_lightning.models.visual_prompters.segment_anything.SegmentAnything.load_state_dict"
         )