refactor(autoware_obect_recognition_utils): rewrite using modern C++ without API breakage

common/autoware_object_recognition_utils/include/autoware/object_recognition_utils/matching.hpp

@@ -31,7 +31,7 @@ namespace autoware::object_recognition_utils
 {
 using autoware_utils_geometry::Polygon2d;
 // minimum area to avoid division by zero
-static const double MIN_AREA = 1e-6;
+static constexpr double MIN_AREA = 1e-6;
 
 inline double getConvexShapeArea(const Polygon2d & source_polygon, const Polygon2d & target_polygon)
 {
@@ -66,7 +66,8 @@ inline double getUnionArea(const Polygon2d & source_polygon, const Polygon2d & t
 }
 
 template <class T1, class T2>
-double get2dIoU(const T1 source_object, const T2 target_object, const double min_union_area = 0.01)
+double get2dIoU(
+  const T1 & source_object, const T2 & target_object, const double min_union_area = 0.01)
 {
   const auto source_polygon = autoware_utils_geometry::to_polygon2d(source_object);
   if (boost::geometry::area(source_polygon) < MIN_AREA) return 0.0;
@@ -99,7 +100,7 @@ double get2dGeneralizedIoU(const T1 & source_object, const T2 & target_object)
 }
 
 template <class T1, class T2>
-double get2dPrecision(const T1 source_object, const T2 target_object)
+double get2dPrecision(const T1 & source_object, const T2 & target_object)
 {
   const auto source_polygon = autoware_utils_geometry::to_polygon2d(source_object);
   const double source_area = boost::geometry::area(source_polygon);
@@ -114,7 +115,7 @@ double get2dPrecision(const T1 source_object, const T2 target_object)
 }
 
 template <class T1, class T2>
-double get2dRecall(const T1 source_object, const T2 target_object)
+double get2dRecall(const T1 & source_object, const T2 & target_object)
 {
   const auto source_polygon = autoware_utils_geometry::to_polygon2d(source_object);
   if (boost::geometry::area(source_polygon) < MIN_AREA) return 0.0;

common/autoware_object_recognition_utils/include/autoware/object_recognition_utils/object_classification.hpp

@@ -15,7 +15,7 @@
 #ifndef AUTOWARE__OBJECT_RECOGNITION_UTILS__OBJECT_CLASSIFICATION_HPP_
 #define AUTOWARE__OBJECT_RECOGNITION_UTILS__OBJECT_CLASSIFICATION_HPP_
 
-#include "autoware_perception_msgs/msg/object_classification.hpp"
+#include <autoware_perception_msgs/msg/object_classification.hpp>
 
 #include <cstdint>
 #include <string>
@@ -39,8 +39,7 @@ inline ObjectClassification getHighestProbClassification(
 inline std::uint8_t getHighestProbLabel(
   const std::vector<ObjectClassification> & object_classifications)
 {
-  auto classification = getHighestProbClassification(object_classifications);
-  return classification.label;
+  return getHighestProbClassification(object_classifications).label;
 }
 
 inline bool isVehicle(const uint8_t label)
@@ -57,8 +56,7 @@ inline bool isVehicle(const ObjectClassification & object_classification)
 
 inline bool isVehicle(const std::vector<ObjectClassification> & object_classifications)
 {
-  auto highest_prob_label = getHighestProbLabel(object_classifications);
-  return isVehicle(highest_prob_label);
+  return isVehicle(getHighestProbLabel(object_classifications));
 }
 
 inline bool isCarLikeVehicle(const uint8_t label)
@@ -74,8 +72,7 @@ inline bool isCarLikeVehicle(const ObjectClassification & object_classification)
 
 inline bool isCarLikeVehicle(const std::vector<ObjectClassification> & object_classifications)
 {
-  auto highest_prob_label = getHighestProbLabel(object_classifications);
-  return isCarLikeVehicle(highest_prob_label);
+  return isCarLikeVehicle(getHighestProbLabel(object_classifications));
 }
 
 inline bool isLargeVehicle(const uint8_t label)
@@ -91,8 +88,7 @@ inline bool isLargeVehicle(const ObjectClassification & object_classification)
 
 inline bool isLargeVehicle(const std::vector<ObjectClassification> & object_classifications)
 {
-  auto highest_prob_label = getHighestProbLabel(object_classifications);
-  return isLargeVehicle(highest_prob_label);
+  return isLargeVehicle(getHighestProbLabel(object_classifications));
 }
 
 inline uint8_t toLabel(const std::string & class_name)
@@ -130,9 +126,7 @@ inline ObjectClassification toObjectClassification(
 inline std::vector<ObjectClassification> toObjectClassifications(
   const std::string & class_name, float probability)
 {
-  std::vector<ObjectClassification> classifications;
-  classifications.push_back(toObjectClassification(class_name, probability));
-  return classifications;
+  return {toObjectClassification(class_name, probability)};
 }
 
 inline std::string convertLabelToString(const uint8_t label)
@@ -158,16 +152,15 @@ inline std::string convertLabelToString(const uint8_t label)
   }
 }
 
-inline std::string convertLabelToString(const ObjectClassification object_classification)
+inline std::string convertLabelToString(const ObjectClassification & object_classification)
 {
   return convertLabelToString(object_classification.label);
 }
 
 inline std::string convertLabelToString(
-  const std::vector<ObjectClassification> object_classifications)
+  const std::vector<ObjectClassification> & object_classifications)
 {
-  auto highest_prob_label = getHighestProbLabel(object_classifications);
-  return convertLabelToString(highest_prob_label);
+  return convertLabelToString(getHighestProbLabel(object_classifications));
 }
 
 }  // namespace autoware::object_recognition_utils

common/autoware_object_recognition_utils/src/conversion.cpp

@@ -43,11 +43,12 @@ DetectedObject toDetectedObject(const TrackedObject & tracked_object)
 
 DetectedObjects toDetectedObjects(const TrackedObjects & tracked_objects)
 {
-  autoware_perception_msgs::msg::DetectedObjects detected_objects;
+  DetectedObjects detected_objects;
   detected_objects.header = tracked_objects.header;
 
+  detected_objects.objects.reserve(tracked_objects.objects.size());
   for (const auto & tracked_object : tracked_objects.objects) {
-    detected_objects.objects.push_back(toDetectedObject(tracked_object));
+    detected_objects.objects.emplace_back(toDetectedObject(tracked_object));
   }
   return detected_objects;
 }