COG: Confidence-aware Optimal Geometric Correspondence for Unsupervised Single-reference Novel Object Pose Estimation

TL;DR

This paper introduces COG, an unsupervised framework for 6DoF object pose estimation that uses confidence-aware optimal transport to find robust correspondences, outperforming existing methods especially with supervision.

Contribution

The paper presents a novel confidence-aware optimal transport approach for correspondence estimation, integrating semantic priors and enabling unsupervised learning for pose estimation.

Findings

Unsupervised COG achieves performance comparable to supervised methods.

Supervised COG surpasses existing approaches in accuracy.

The method effectively handles occlusions and view-point changes.

Abstract

Estimating the 6DoF pose of a novel object with a single reference view is challenging due to occlusions, view-point changes, and outliers. A core difficulty lies in finding robust cross-view correspondences, as existing methods often rely on discrete one-to-one matching that is non-differentiable and tends to collapse onto sparse key-points. We propose Confidence-aware Optimal Geometric Correspondence (COG), an unsupervised framework that formulates correspondence estimation as a confidence-aware optimal transport problem. COG produces balanced soft correspondences by predicting point-wise confidences and injecting them as optimal transport marginals, suppressing non-overlapping regions. Semantic priors from vision foundation models further regularize the correspondences, leading to stable pose estimation. This design integrates confidence into the correspondence finding and pose estimation pipeline, enabling unsupervised learning. Experiments show unsupervised COG achieves comparable performance to supervised methods, and supervised COG outperforms them.