[CVAT] Updates (#3445)

Author: zhiltsov-max

packages/examples/cvat/exchange-oracle/src/core/tasks/boxes_from_points.py

@@ -8,6 +8,8 @@
 from attrs import frozen
 from datumaro.util import dump_json, parse_json
 
+from src.core.tasks.roi_based import OUTPUT_OBJECT_ID_ATTR
+
 BboxPointMapping = dict[int, int]
 
 
@@ -90,3 +92,14 @@ def parse_roi_info(self, rois_info_data: bytes) -> RoiInfos:
 
     def parse_roi_filenames(self, roi_filenames_data: bytes) -> RoiFilenames:
         return {int(k): v for k, v in parse_json(roi_filenames_data).items()}
+
+
+__all__ = [
+    "OUTPUT_OBJECT_ID_ATTR",
+    "BboxPointMapping",
+    "RoiInfo",
+    "RoiInfos",
+    "RoiFilenames",
+    "TaskMetaLayout",
+    "TaskMetaSerializer",
+]

packages/examples/cvat/exchange-oracle/src/core/tasks/roi_based.py

@@ -0,0 +1 @@
+OUTPUT_OBJECT_ID_ATTR = "object_id"

packages/examples/cvat/exchange-oracle/src/core/tasks/skeletons_from_boxes.py

@@ -9,6 +9,7 @@
 from datumaro.util import dump_json, parse_json
 
 from src.core.config import Config
+from src.core.tasks.roi_based import OUTPUT_OBJECT_ID_ATTR
 
 DEFAULT_ASSIGNMENT_SIZE_MULTIPLIER = Config.core_config.skeleton_assignment_size_mult
 
@@ -133,3 +134,15 @@ def parse_point_labels(self, point_labels_data: bytes) -> PointLabelsMapping:
             (v["skeleton_label"], v["point_label"]): v["job_point_label"]
             for v in parse_json(point_labels_data)
         }
+
+
+__all__ = [
+    "OUTPUT_OBJECT_ID_ATTR",
+    "SkeletonBboxMapping",
+    "RoiInfo",
+    "RoiInfos",
+    "RoiFilenames",
+    "PointLabelsMapping",
+    "TaskMetaLayout",
+    "TaskMetaSerializer",
+]

packages/examples/cvat/exchange-oracle/src/handlers/job_export.py

@@ -262,6 +262,18 @@ def _process_merged_dataset(self, input_dataset: Dataset) -> Dataset:
             ).elements[0]
             old_x, old_y = old_point.points[:2]
 
+            roi_anns = [
+                roi_ann.wrap(
+                    id=roi_info.point_id,
+                    attributes={
+                        **roi_sample.attributes,
+                        boxes_from_points_task.OUTPUT_OBJECT_ID_ATTR: roi_info.point_id,
+                    },
+                )
+                for roi_ann in roi_sample.annotations
+                if isinstance(roi_ann, dm.Bbox)
+            ]
+
             merged_sample.annotations.extend(
                 annotation_utils.shift_ann(
                     roi_ann,
@@ -270,8 +282,7 @@ def _process_merged_dataset(self, input_dataset: Dataset) -> Dataset:
                     img_w=image_w,
                     img_h=image_h,
                 )
-                for roi_ann in roi_sample.annotations
-                if isinstance(roi_ann, dm.Bbox)
+                for roi_ann in roi_anns
             )
 
         return merged_sample_dataset
@@ -456,6 +467,7 @@ def _process_dataset(self, dataset: Dataset, *, ann_descriptor: FileDescriptor)
                     ],
                     label=self.bbox_label_to_merged[old_bbox.label],
                     id=old_bbox.id,
+                    attributes={skeletons_from_boxes_task.OUTPUT_OBJECT_ID_ATTR: old_bbox.id},
                 )
                 skeleton_sample.annotations.append(merged_skeleton)
 

packages/examples/cvat/recording-oracle/src/core/config.py

@@ -177,6 +177,12 @@ class ValidationConfig:
     value small enough for faster convergence rate of the annotation process.
     """
 
+    enable_gt_bans = to_bool(getenv("ENABLE_GT_BANS", "1"))
+    """
+    Whether to allow automatic GT bans for bad images or not. By default, bans are allowed.
+    This can raise escrow annotation chances at the cost of reduced quality threshold.
+    """
+
     unverifiable_assignments_threshold = float(getenv("UNVERIFIABLE_ASSIGNMENTS_THRESHOLD", "0.1"))
     """
     Deprecated. Not expected to happen in practice, kept only as a safety fallback rule.

packages/examples/cvat/recording-oracle/src/handlers/process_intermediate_results.py

@@ -574,7 +574,8 @@ def _get_available_gt(self) -> tuple[GtStats, dict[int, set[GtKey]]]:
             gt_key
             for gt_key, gt_stat in self.gt_stats.items()
             if gt_stat.enabled
-            if gt_stat.rating > 1 - self.manifest.validation.min_quality
+            if not Config.validation.enable_gt_bans
+            or (gt_stat.rating > 1 - self.manifest.validation.min_quality)
         }, task_id_to_gt_keys
 
     def _check_warmup_annotation_speed(self):