PCRP: Unsupervised Point Cloud Object Retrieval and Pose Estimation

TL;DR

PCRP is an unsupervised method for point cloud object retrieval and pose estimation that registers unknown objects with a gallery set, outperforming traditional and learning-based approaches on ModelNet40.

Contribution

It introduces PCRP, a novel unsupervised approach combining object retrieval and pose estimation through enhanced point cloud registration.

Findings

PCRP outperforms traditional methods on ModelNet40.

PCRP achieves accurate pose estimation without supervision.

The method effectively combines retrieval and registration tasks.

Abstract

An unsupervised point cloud object retrieval and pose estimation method, called PCRP, is proposed in this work. It is assumed that there exists a gallery point cloud set that contains point cloud objects with given pose orientation information. PCRP attempts to register the unknown point cloud object with those in the gallery set so as to achieve content-based object retrieval and pose estimation jointly, where the point cloud registration task is built upon an enhanced version of the unsupervised R-PointHop method. Experiments on the ModelNet40 dataset demonstrate the superior performance of PCRP in comparison with traditional and learning based methods.