Mergeback releases/1.5.0 to develop (#2830)

* Update MAPI version (#2730)

* Update dependency for exportable code (#2732)

* Filter invalid polygon shapes (#2795)

---------

Co-authored-by: Vladislav Sovrasov <sovrasov.vlad@gmail.com>
Co-authored-by: Eugene Liu <eugene.liu@intel.com>

requirements/openvino.txt

@@ -2,7 +2,7 @@
 # OpenVINO Requirements.                                                      #
 nncf==2.7.0
 onnx==1.13.0
-openvino-model-api==0.1.7
+openvino-model-api==0.1.8
 openvino==2023.2.0
 openvino-dev==2023.2.0
 openvino-telemetry==2023.2.*

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

@@ -188,27 +188,30 @@ def infer(self, dataset: DatasetEntity, inference_parameters: InferenceParameter
                 label = self.anomalous_label if image_result.pred_score >= 0.5 else self.normal_label
             elif self.task_type == TaskType.ANOMALY_SEGMENTATION:
                 annotations = create_annotation_from_segmentation_map(
-                    pred_mask, image_result.anomaly_map.squeeze(), {0: self.normal_label, 1: self.anomalous_label}
+                    pred_mask,
+                    image_result.anomaly_map.squeeze() / 255.0,
+                    {0: self.normal_label, 1: self.anomalous_label},
                 )
                 dataset_item.append_annotations(annotations)
                 label = self.normal_label if len(annotations) == 0 else self.anomalous_label
             elif self.task_type == TaskType.ANOMALY_DETECTION:
                 annotations = create_detection_annotation_from_anomaly_heatmap(
-                    pred_mask, image_result.anomaly_map.squeeze(), {0: self.normal_label, 1: self.anomalous_label}
+                    pred_mask,
+                    image_result.anomaly_map.squeeze() / 255.0,
+                    {0: self.normal_label, 1: self.anomalous_label},
                 )
                 dataset_item.append_annotations(annotations)
                 label = self.normal_label if len(annotations) == 0 else self.anomalous_label
             else:
                 raise ValueError(f"Unknown task type: {self.task_type}")
 
             dataset_item.append_labels([ScoredLabel(label=label, probability=float(probability))])
-            anomaly_map = (image_result.anomaly_map * 255).astype(np.uint8)
             heatmap_media = ResultMediaEntity(
                 name="Anomaly Map",
                 type="anomaly_map",
                 label=label,
                 annotation_scene=dataset_item.annotation_scene,
-                numpy=anomaly_map,
+                numpy=image_result.anomaly_map,
             )
             dataset_item.append_metadata_item(heatmap_media)
             update_progress_callback(int((idx + 1) / len(dataset) * 100))

src/otx/api/usecases/exportable_code/demo/requirements.txt

@@ -1,4 +1,4 @@
 openvino==2023.2.0
-openvino-model-api==0.1.7
+openvino-model-api==0.1.8
 otx @ git+https://github.com/openvinotoolkit/training_extensions/@2988fdc51ef7e4a136a9d4e09602b3844d7bafec#egg=otx
 numpy>=1.21.0,<=1.23.5  # np.bool was removed in 1.24.0 which was used in openvino runtime

src/otx/api/usecases/exportable_code/prediction_to_annotation_converter.py

@@ -380,7 +380,7 @@ def convert_to_annotation(self, predictions: AnomalyResult, metadata: Dict[str,
         assert predictions.pred_mask is not None
         assert predictions.anomaly_map is not None
         annotations = create_annotation_from_segmentation_map(
-            predictions.pred_mask, predictions.anomaly_map, self.label_map
+            predictions.pred_mask, predictions.anomaly_map / 255.0, self.label_map
         )
         if len(annotations) == 0:
             # TODO: add confidence to this label

src/otx/core/data/adapter/base_dataset_adapter.py

@@ -273,11 +273,16 @@ def _prepare_label_information(
 
         return {"category_items": category_items, "label_groups": label_groups, "label_entities": label_entities}
 
-    def _is_normal_polygon(self, annotation: DatumAnnotationType.polygon) -> bool:
+    def _is_normal_polygon(self, annotation: DatumAnnotationType.polygon, width: int, height: int) -> bool:
         """To filter out the abnormal polygon."""
-        x_points = [annotation.points[i] for i in range(0, len(annotation.points), 2)]
-        y_points = [annotation.points[i + 1] for i in range(0, len(annotation.points), 2)]
-        return min(x_points) < max(x_points) and min(y_points) < max(y_points)
+        x_points = annotation.points[::2]  # Extract x-coordinates
+        y_points = annotation.points[1::2]  # Extract y-coordinates
+
+        return (
+            min(x_points) < max(x_points) < width
+            and min(y_points) < max(y_points) < height
+            and annotation.get_area() > 0
+        )
 
     def _is_normal_bbox(self, x1: float, y1: float, x2: float, y2: float) -> bool:
         """To filter out the abrnormal bbox."""

src/otx/core/data/adapter/detection_dataset_adapter.py

@@ -37,8 +37,11 @@ def get_otx_dataset(self) -> DatasetEntity:
                     assert isinstance(image, Image)
                     shapes = []
                     for ann in datumaro_item.annotations:
-                        if self.task_type in (TaskType.INSTANCE_SEGMENTATION, TaskType.ROTATED_DETECTION):
-                            if ann.type == DatumAnnotationType.polygon and self._is_normal_polygon(ann):
+                        if (
+                            self.task_type in (TaskType.INSTANCE_SEGMENTATION, TaskType.ROTATED_DETECTION)
+                            and ann.type == DatumAnnotationType.polygon
+                        ):
+                            if self._is_normal_polygon(ann, image.width, image.height):
                                 shapes.append(self._get_polygon_entity(ann, image.width, image.height))
                             elif ann.type == DatumAnnotationType.ellipse:
                                 shapes.append(self._get_ellipse_entity(ann, image.width, image.height))

src/otx/core/data/adapter/visual_prompting_dataset_adapter.py

@@ -53,7 +53,7 @@ def get_otx_dataset(self) -> DatasetEntity:
                     for ann in datumaro_item.annotations:
                         if ann.type == DatumAnnotationType.polygon:
                             # save polygons as-is, they will be converted to masks.
-                            if self._is_normal_polygon(ann):
+                            if self._is_normal_polygon(ann, image.width, image.height):
                                 shapes.append(self._get_polygon_entity(ann, image.width, image.height))
 
                         if ann.type == DatumAnnotationType.mask:

tests/unit/algorithms/segmentation/adapters/mmseg/datasets/pipelines/test_transforms.py

@@ -110,8 +110,8 @@ class TestNormalize:
     @pytest.mark.parametrize(
         "mean,std,to_rgb,expected",
         [
-            ([1.0 for _ in range(3)], [1.0 for _ in range(3)], True, np.array([[[1.0, 0.0, -1.0]]], dtype=np.float32)),
-            ([1.0 for _ in range(3)], [1.0 for _ in range(3)], False, np.array([[[-1.0, 0.0, 1.0]]], dtype=np.float32)),
+            ([[[1.0, 1.0, 1.0]]], [[[1.0, 1.0, 1.0]]], True, np.array([[[1.0, 0.0, -1.0]]], dtype=np.float32)),
+            ([[[1.0, 1.0, 1.0]]], [[[1.0, 1.0, 1.0]]], False, np.array([[[-1.0, 0.0, 1.0]]], dtype=np.float32)),
         ],
     )
     def test_call(self, mean: List[float], std: List[float], to_rgb: bool, expected: np.array) -> None: