Reduce code duplication

cvat/apps/consensus/intersect_merge.py

@@ -7,7 +7,7 @@
 import itertools
 from abc import ABCMeta, abstractmethod
 from collections.abc import Collection
-from typing import Iterable, Optional, Sequence
+from typing import Iterable, Sequence
 
 import attrs
 
@@ -17,7 +17,7 @@
 import datumaro.components.merge.intersect_merge
 import numpy as np
 from datumaro.components.errors import FailedLabelVotingError
-from datumaro.util.annotation_util import find_instances, max_bbox, mean_bbox
+from datumaro.util.annotation_util import mean_bbox
 from datumaro.util.attrs_util import ensure_cls
 
 from cvat.apps.engine.models import Label
@@ -121,27 +121,8 @@ def _for_type(t: dm.AnnotationType, **kwargs) -> AnnotationMatcher:
             else:
                 raise AssertionError(f"Annotation type {t} is not supported")
 
-        instance_map = {}
-        for s in sources:
-            s_instances = find_instances(s)
-            for inst in s_instances:
-                inst_bbox = max_bbox(
-                    [
-                        a
-                        for a in inst
-                        if a.type
-                        in {
-                            dm.AnnotationType.polygon,
-                            dm.AnnotationType.mask,
-                            dm.AnnotationType.bbox,
-                        }
-                    ]
-                )
-                for ann in inst:
-                    instance_map[id(ann)] = [inst, inst_bbox]
-
         self._mergers = {
-            t: _for_type(t, instance_map=instance_map, categories=self._categories)
+            t: _for_type(t, categories=self._categories)
             for t in self.conf.included_annotation_types
         }
 
@@ -357,17 +338,13 @@ class MaskMatcher(PolygonMatcher):
 
 @attrs.define(kw_only=True, slots=False)
 class PointsMatcher(ShapeMatcher):
-    sigma: Optional[list] = attrs.field(default=None)
-
     def match_annotations_between_two_items(self, item_a, item_b):
         matches, _, _, _, distances = self._comparator.match_points(item_a, item_b)
         return matches, distances
 
 
 @attrs.define(kw_only=True, slots=False)
 class SkeletonMatcher(ShapeMatcher):
-    sigma: float = 0.1
-
     def match_annotations_between_two_items(self, item_a, item_b):
         matches, _, _, _, distances = self._comparator.match_skeletons(item_a, item_b)
         return matches, distances