robokudo.annotators.icp_pose_refinement

Classes

ICPPoseRefinementAnnotator

Read existing Pose Annotations from a Cluster and try to improve them by applying ICP.

Module Contents

class robokudo.annotators.icp_pose_refinement.ICPPoseRefinementAnnotator(name='ICPPoseRefinementAnnotator', descriptor=Descriptor())

Bases: robokudo.annotators.core.ThreadedAnnotator

Read existing Pose Annotations from a Cluster and try to improve them by applying ICP. The model for ICP has to be stated explicitly in the parameters.

Please note, that of today we can only refine against one model that has been loaded in the beginning. self.class_name_to_ply_model could be extended in the future to cope with this limitation.

PointCloud models can be easily generated from meshes by using the tool CloudCompare. Check out the “Sample points on a mesh” feature.

class Descriptor

Bases: robokudo.annotators.core.BaseAnnotator.Descriptor

class Parameters
ros_pkg_path = None
data_path = None
parameters
file_loader
class_name_to_ply_model
get_model_path(class_name: str = '')
compute()

This method is doing the heavy lifting of the annotator, if it takes to long to run.