robokudo.annotators.object_knowledge_visualizer

Object knowledge visualization for RoboKudo.

This module provides an annotator for visualizing object knowledge, including reference frames, components, and features of detected objects. It integrates with the object knowledge base to display semantic information about objects and their parts.

Classes

ObjectKnowledgeVisualizer

Annotator for visualizing object knowledge and part relationships.

Module Contents

class robokudo.annotators.object_knowledge_visualizer.ObjectKnowledgeVisualizer(name='ObjectKnowledgeVisualizer', descriptor=Descriptor())

Bases: robokudo.annotators.core.BaseAnnotator

Annotator for visualizing object knowledge and part relationships.

This annotator displays reference frames of objects and their components/features by integrating with the object knowledge base. It creates visualizations showing: - Object reference frames - Component and feature bounding boxes - Part relationships - 2D ROIs for parts in images

Variables:

object_kb – Object knowledge base instance

class Descriptor

Bases: robokudo.annotators.core.BaseAnnotator.Descriptor

Configuration descriptor for object knowledge visualization.

class Parameters

Parameters for configuring object knowledge visualization.

Variables:

  • object_knowledge_base_ros_package – ROS package containing knowledge base, defaults to “robokudo”

  • object_knowledge_base_name – Name of knowledge base module, defaults to “object_knowledge_iai_kitchen”

object_knowledge_base_ros_package = 'robokudo'
object_knowledge_base_name = 'object_knowledge_iai_kitchen'
parameters
object_kb
fill_parthood_hypothesis(ph: robokudo.types.scene.ParthoodHypothesis, object_knowledge: robokudo.descriptors.object_knowledge.object_knowledge_iai_kitchen.ObjectKnowledge, transform: numpy.typing.NDArray) → bool

Insert into a given ParthoodHypothesis the information that can be extracted from the object knowledge

Parameters:

  • ph (ParthoodHypothesis) – A ParthoodHypothesis or one of its childs.

  • object_knowledge (ObjectKnowledge) – Knowledge about the object and its parts

  • transform (npt.NDArray) – Transform between cam and parent object

Returns:

bool: True if parthood hypothesis could be generated, False otherwise

Return type:

bool

generate_parthood_hypotheses(object_knowledge: robokudo.descriptors.object_knowledge.object_knowledge_iai_kitchen.ObjectKnowledge, transform: numpy.typing.NDArray)

Generate hypotheses for object parts and features.

Creates ParthoodComponentHypothesis and ParthoodFeatureHypothesis objects for all components and features defined in the object knowledge.

Parameters:

  • object_knowledge (ObjectKnowledge) – Knowledge about the object and its parts

  • transform (npt.NDArray) – Transform between camera and object

Returns:

List of generated hypotheses or None if no parts exist

Return type:

list[ParthoodHypothesis] or None

update()

Update the visualization with current object knowledge.

Creates visualizations containing:

  • Object reference frames

  • Oriented bounding boxes for objects and parts

  • 2D ROIs in the color image

  • Part relationship annotations in the CAS

Returns:

SUCCESS after creating visualizations

Return type:

py_trees.common.Status