Add tests for py bindings

Author: sovrasov

.github/workflows/test_accuracy.yml

@@ -49,3 +49,7 @@ jobs:
         run: |
           build/test_accuracy -d data -p tests/python/accuracy/public_scope.json
           DATA=data build/test_YOLOv8
+      - name: Run CPP-PY Bindings Test
+        run: |
+          source venv/bin/activate
+          PYTHONPATH="$PYTHONPATH:build/py_bindings" pytest --data=./data --config=./tests/python/accuracy/public_scope.json tests/cpp/accuracy/test_bindings.py

tests/cpp/accuracy/conftest.py

@@ -0,0 +1,8 @@
+#
+# Copyright (C) 2025 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+#
+
+def pytest_addoption(parser):
+    parser.addoption("--data", action="store", help="data folder with dataset")
+    parser.addoption("--config", action="store", help="path to models config")

tests/cpp/accuracy/test_bindings.py

@@ -0,0 +1,56 @@
+#
+# Copyright (C) 2025 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+#
+
+import pytest
+import json
+from pathlib import Path
+
+import cv2
+
+from model_api.models import Model
+from py_model_api import ClassificationModel
+
+
+def read_config(models_config: str, model_type: str):
+    with open(models_config, "r") as f:
+        data = json.load(f)
+        for item in data:
+            if item["type"] == model_type:
+                yield item
+
+
+@pytest.fixture(scope="session")
+def data(pytestconfig) -> str:
+    return pytestconfig.getoption("data")
+
+
+@pytest.fixture(scope="session")
+def models_config(pytestconfig) -> str:
+    return pytestconfig.getoption("config")
+
+
+@pytest.fixture()
+def classification_configs(models_config: str):
+    return read_config(models_config, "ClassificationModel")
+
+
+def test_classification_models(data: str, classification_configs):
+    for model_data in classification_configs:
+        name = model_data["name"]
+        if ".xml" not in name:
+            continue
+        if name.endswith(".xml") or name.endswith(".onnx"):
+            name = f"{data}/{name}"
+
+        model = Model.create_model(name, preload=True)
+        cpp_model = ClassificationModel.create_model(name, preload=True)
+
+        image_path = Path(data) / next(iter(model_data["test_data"]))["image"]
+        image = cv2.imread(str(image_path))
+
+        py_result = model(image)
+        cpp_result = cpp_model(image)
+
+        assert str(py_result) == str(cpp_result)