Cleanup

Signed-off-by: Albert van Houten <albert.van.houten@intel.com>

Author: AlbertvanHouten

library/src/otx/backend/native/models/detection/ssd.py

@@ -15,7 +15,6 @@
 from typing import TYPE_CHECKING, Any, ClassVar, Literal
 
 import numpy as np
-from datumaro.experimental.dataset import Dataset as DmDataset
 
 from otx.backend.native.exporter.base import OTXModelExporter
 from otx.backend.native.exporter.native import OTXNativeModelExporter
@@ -29,7 +28,6 @@
 from otx.backend.native.models.detection.utils.prior_generators import SSDAnchorGeneratorClustered
 from otx.backend.native.models.utils.utils import load_checkpoint
 from otx.config.data import TileConfig
-from otx.data.entity.sample import DetectionSample
 from otx.metrics.fmeasure import MeanAveragePrecisionFMeasureCallable
 
 if TYPE_CHECKING:
@@ -245,19 +243,7 @@ def _get_sizes_from_dataset_entity(dataset: OTXDataset, target_wh: list[int]) ->
             list[tuple[int, int]]: tuples with width and height of each instance
         """
         wh_stats = np.empty((0, 2), dtype=np.float32)
-        if not isinstance(dataset.dm_subset, DmDataset):
-            exc_str = "The variable dataset.dm_subset must be an instance of DmDataset"
-            raise TypeError(exc_str)
-
         for item in dataset.dm_subset:
-            if not isinstance(item, DetectionSample):
-                exc_str = "The variable item must be an instance of DetectionSample"
-                raise TypeError(exc_str)
-
-            if item.img_info is None:
-                exc_str = "The image info must not be None"
-                raise RuntimeError(exc_str)
-
             height, width = item.img_info.img_shape
             x1 = item.bboxes[:, 0]
             y1 = item.bboxes[:, 1]

library/src/otx/data/entity/tile.py

@@ -19,7 +19,7 @@
 from .base import ImageInfo
 
 if TYPE_CHECKING:
-    from datumaro import Polygon
+    import numpy as np
     from datumaro.experimental.fields import TileInfo
     from torch import LongTensor
 
@@ -162,13 +162,13 @@ class TileInstSegDataEntity(TileDataEntity):
         ori_bboxes (tv_tensors.BoundingBoxes): The bounding boxes of the original image.
         ori_labels (LongTensor): The labels of the original image.
         ori_masks (tv_tensors.Mask): The masks of the original image.
-        ori_polygons (list[Polygon]): The polygons of the original image.
+        ori_polygons (list[np.ndarray]): The polygons of the original image as arrays shaped (K, 2).
     """
 
     ori_bboxes: tv_tensors.BoundingBoxes
     ori_labels: LongTensor
     ori_masks: tv_tensors.Mask
-    ori_polygons: list[Polygon]
+    ori_polygons: list[np.ndarray]
 
     @property
     def task(self) -> OTXTaskType:
@@ -184,13 +184,13 @@ class TileBatchInstSegDataEntity(OTXTileBatchDataEntity):
         bboxes (list[tv_tensors.BoundingBoxes]): The bounding boxes of the original image.
         labels (list[LongTensor]): The labels of the original image.
         masks (list[tv_tensors.Mask]): The masks of the original image.
-        polygons (list[list[Polygon]]): The polygons of the original image.
+        polygons (list[list[np.ndarray]]): The polygons of the original image as arrays shaped (K, 2).
     """
 
     bboxes: list[tv_tensors.BoundingBoxes]
     labels: list[LongTensor]
     masks: list[tv_tensors.Mask]
-    polygons: list[list[Polygon]]
+    polygons: list[list[np.ndarray]]
 
     def unbind(self) -> list[tuple[TileAttrDictList, OTXDataBatch]]:
         """Unbind batch data entity for instance segmentation task."""

library/src/otx/data/factory.py

@@ -7,11 +7,8 @@
 
 from typing import TYPE_CHECKING
 
-from datumaro.components.annotation import AnnotationType
-from datumaro.experimental.categories import LabelCategories
 from datumaro.experimental.legacy import convert_from_legacy
 
-from otx import LabelInfo, NullLabelInfo
 from otx.types.task import OTXTaskType
 from otx.types.transformer_libs import TransformLibType
 
@@ -115,13 +112,3 @@ def create(
             return OTXKeypointDetectionDataset(**common_kwargs)
 
         raise NotImplementedError(task)
-
-    @staticmethod
-    def _get_label_categories(dm_subset: DmDataset, data_format: str) -> LabelCategories:
-        if dm_subset.categories() and data_format == "arrow":
-            label_info = LabelInfo.from_dm_label_groups_arrow(dm_subset.categories()[AnnotationType.label])
-        elif dm_subset.categories():
-            label_info = LabelInfo.from_dm_label_groups(dm_subset.categories()[AnnotationType.label])
-        else:
-            label_info = NullLabelInfo()
-        return LabelCategories(labels=label_info.label_names)

library/src/otx/data/transform_libs/torchvision.py

@@ -1429,7 +1429,7 @@ def _transform_polygons(
         filtered_polygons = (
             [p for p, keep in zip(inputs.polygons, valid_index) if keep] if valid_index is not None else inputs.polygons
         )
-        if len(filtered_polygons) > 0:
+        if filtered_polygons:
             inputs.polygons = project_polygons(filtered_polygons, warp_matrix, output_shape)
 
     def _recompute_bboxes(self, inputs: OTXDataItem, output_shape: tuple[int, int]) -> None:

library/tests/unit/tools/test_converter.py

@@ -112,7 +112,6 @@ def test_classification_augs(self, tmp_path):
         assert engine.datamodule.train_dataloader().dataset.transforms is not None
         assert len(engine.datamodule.train_dataloader().dataset.transforms.transforms) == 9
 
-    @pytest.mark.xfail(reason="Tiling not yet implemented with new dataset")
     def test_detection_augs(self, tmp_path):
         supported_augs_list_for_configuration = [
             "otx.data.transform_libs.torchvision.MinIoURandomCrop",