Confronting Ambiguity in 6D Object Pose Estimation via Score-Based Diffusion on SE(3)

TL;DR

This paper introduces a novel score-based diffusion approach on SE(3) for 6D object pose estimation from RGB images, effectively addressing pose ambiguity, symmetry, and occlusion challenges with improved efficiency.

Contribution

We present the first application of diffusion models on SE(3) for pose estimation, introducing a surrogate Stein score formulation that enhances convergence and computational efficiency.

Findings

Demonstrates robustness in handling pose ambiguity and occlusions

Improves convergence and efficiency of pose estimation process

Validates effectiveness through extensive evaluations

Abstract

Addressing pose ambiguity in 6D object pose estimation from single RGB images presents a significant challenge, particularly due to object symmetries or occlusions. In response, we introduce a novel score-based diffusion method applied to the $SE(3)$ group, marking the first application of diffusion models to $SE(3)$ within the image domain, specifically tailored for pose estimation tasks. Extensive evaluations demonstrate the method's efficacy in handling pose ambiguity, mitigating perspective-induced ambiguity, and showcasing the robustness of our surrogate Stein score formulation on $SE(3)$. This formulation not only improves the convergence of denoising process but also enhances computational efficiency. Thus, we pioneer a promising strategy for 6D object pose estimation.