feat: update PerceptionFrameConfig

Signed-off-by: ktro2828 <[email protected]>

perception_eval/perception_eval/manager/perception_evaluation_manager.py

@@ -79,8 +79,8 @@ def add_frame_result(
         unix_time: int,
         ground_truth_now_frame: FrameGroundTruth,
         estimated_objects: List[ObjectType],
-        critical_ground_truth_objects: Optional[List[ObjectType]] = None,
         frame_config: Optional[PerceptionFrameConfig] = None,
+        critical_ground_truth_objects: Optional[List[ObjectType]] = None,
     ) -> PerceptionFrameResult:
         """Get perception result at current frame.
 
@@ -111,7 +111,7 @@ def add_frame_result(
             critical_ground_truth_objects = ground_truth_now_frame.objects.copy()
 
         if frame_config is None:
-            frame_config = PerceptionFrameConfig(self.config)
+            frame_config = PerceptionFrameConfig.from_eval_cfg(self.config)
 
         result = PerceptionFrameResult(
             unix_time=unix_time,

perception_eval/perception_eval/result/perception/perception_frame_config.py

@@ -14,16 +14,17 @@
 
 from __future__ import annotations
 
-from typing import Any
-from typing import Dict
 from typing import List
 from typing import Optional
 from typing import TYPE_CHECKING
+from typing import Union
 
 from perception_eval.common.label import set_target_lists
 from perception_eval.common.threshold import check_thresholds
+from perception_eval.config.params import PerceptionFilterParam
 
 if TYPE_CHECKING:
+    from perception_eval.common.label import LabelType
     from perception_eval.config import PerceptionEvaluationConfig
 
 
@@ -52,15 +53,15 @@ class PerceptionFrameConfig:
     def __init__(
         self,
         evaluator_config: PerceptionEvaluationConfig,
-        target_labels: List[str],
-        ignore_attributes: Optional[List[str]] = None,
+        target_labels: List[Union[str, LabelType]],
         max_x_position_list: Optional[List[float]] = None,
         max_y_position_list: Optional[List[float]] = None,
-        max_distance_list: Optional[List[float]] = None,
         min_distance_list: Optional[List[float]] = None,
+        max_distance_list: Optional[List[float]] = None,
         min_point_numbers: Optional[List[int]] = None,
         confidence_threshold_list: Optional[List[float]] = None,
         target_uuids: Optional[List[str]] = None,
+        ignore_attributes: Optional[List[str]] = None,
         thresholds: Optional[List[float]] = None,
     ) -> None:
         """
@@ -82,56 +83,42 @@ def __init__(
             target_uuids (Optional[List[str]]): The list of target uuid. Defaults to None.
         """
         self.evaluation_task = evaluator_config.evaluation_task
-
-        self.target_labels = set_target_lists(target_labels, evaluator_config.label_converter)
-        self.ignore_attributes = ignore_attributes
-
-        num_elements: int = len(self.target_labels)
-        if max_x_position_list and max_y_position_list:
-            self.max_x_position_list: List[float] = check_thresholds(max_x_position_list, num_elements)
-            self.max_y_position_list: List[float] = check_thresholds(max_y_position_list, num_elements)
-            self.max_distance_list = None
-            self.min_distance_list = None
-        elif max_distance_list and min_distance_list:
-            self.max_distance_list: List[float] = check_thresholds(max_distance_list, num_elements)
-            self.min_distance_list: List[float] = check_thresholds(min_distance_list, num_elements)
-            self.max_x_position_list = None
-            self.max_y_position_list = None
-        elif evaluator_config.evaluation_task.is_2d():
-            self.max_x_position_list = None
-            self.max_y_position_list = None
-            self.max_distance_list = None
-            self.min_distance_list = None
+        if all([isinstance(label, str) for label in target_labels]):
+            self.target_labels = set_target_lists(target_labels, evaluator_config.label_converter)
         else:
-            raise RuntimeError("Either max x/y position or max/min distance should be specified")
-
-        if min_point_numbers is None:
-            self.min_point_numbers = None
-        else:
-            self.min_point_numbers: List[int] = check_thresholds(min_point_numbers, num_elements)
-
-        if confidence_threshold_list is None:
-            self.confidence_threshold_list = None
-        else:
-            self.confidence_threshold_list: List[float] = check_thresholds(confidence_threshold_list, num_elements)
-
-        self.target_uuids: Optional[List[str]] = target_uuids
-
-        self.filtering_params: Dict[str, Any] = {
-            "target_labels": self.target_labels,
-            "ignore_attributes": self.ignore_attributes,
-            "max_x_position_list": self.max_x_position_list,
-            "max_y_position_list": self.max_y_position_list,
-            "max_distance_list": self.max_distance_list,
-            "min_distance_list": self.min_distance_list,
-            "min_point_numbers": self.min_point_numbers,
-            "confidence_threshold_list": self.confidence_threshold_list,
-            "target_uuids": self.target_uuids,
-        }
+            self.target_labels = target_labels
+
+        self.filter_param = PerceptionFilterParam(
+            evaluation_task=self.evaluation_task,
+            target_labels=self.target_labels,
+            max_x_position_list=max_x_position_list,
+            max_y_position_list=max_y_position_list,
+            min_distance_list=min_distance_list,
+            max_distance_list=max_distance_list,
+            min_point_numbers=min_point_numbers,
+            confidence_threshold_list=confidence_threshold_list,
+            target_uuids=target_uuids,
+            ignore_attributes=ignore_attributes,
+        )
 
         num_elements: int = len(self.target_labels)
-
         if thresholds is None:
             self.thresholds = None
         else:
             self.thresholds = check_thresholds(thresholds, num_elements)
+
+    @classmethod
+    def from_eval_cfg(cls, eval_cfg: PerceptionEvaluationConfig) -> PerceptionFrameConfig:
+        return cls(
+            evaluator_config=eval_cfg,
+            target_labels=eval_cfg.target_labels,
+            max_x_position_list=eval_cfg.filter_param.max_x_position_list,
+            max_y_position_list=eval_cfg.filter_param.max_y_position_list,
+            min_distance_list=eval_cfg.filter_param.min_distance_list,
+            max_distance_list=eval_cfg.filter_param.max_distance_list,
+            min_point_numbers=eval_cfg.filter_param.min_point_numbers,
+            confidence_threshold_list=eval_cfg.filter_param.confidence_threshold_list,
+            target_uuids=eval_cfg.filter_param.target_uuids,
+            ignore_attributes=eval_cfg.filter_param.ignore_attributes,
+            thresholds=eval_cfg.metrics_param.plane_distance_thresholds,
+        )

perception_eval/perception_eval/result/perception/perception_pass_fail_result.py

@@ -84,7 +84,7 @@ def evaluate(
             objects=critical_ground_truth_objects,
             is_gt=True,
             ego2map=self.ego2map,
-            **self.frame_config.filtering_params,
+            **self.frame_config.filter_param.as_dict(),
         )
         self.tp_object_results, self.fp_object_results = self.__get_positive_object_results(
             object_results=object_results,

perception_eval/test/perception_fp_validation_lsim.py

@@ -82,8 +82,8 @@ def callback(self, unix_time: int, estimated_objects: List[ObjectType]) -> None:
             unix_time=unix_time,
             ground_truth_now_frame=ground_truth_now_frame,
             estimated_objects=estimated_objects,
-            critical_ground_truth_objects=critical_ground_truth_objects,
             frame_config=frame_config,
+            critical_ground_truth_objects=critical_ground_truth_objects,
         )
         self.display(frame_result)
 

perception_eval/test/perception_lsim.py

@@ -123,8 +123,8 @@ def callback(
             unix_time=unix_time,
             ground_truth_now_frame=ground_truth_now_frame,
             estimated_objects=estimated_objects,
-            critical_ground_truth_objects=critical_ground_truth_objects,
             frame_config=frame_config,
+            critical_ground_truth_objects=critical_ground_truth_objects,
         )
         self.visualize(frame_result)
 

perception_eval/test/perception_lsim2d.py

@@ -129,8 +129,8 @@ def callback(
             unix_time=unix_time,
             ground_truth_now_frame=ground_truth_now_frame,
             estimated_objects=estimated_objects,
-            critical_ground_truth_objects=critical_ground_truth_objects,
             frame_config=frame_config,
+            critical_ground_truth_objects=critical_ground_truth_objects,
         )
         self.visualize(frame_result)