OneViewAll: Semantic Prior Guided One-View 6D Pose Estimation for Novel Objects

2026-05-07

AI Summary

This work proposes OneViewAll, a novel framework for 6D object pose estimation of previously unseen objects using only a single real RGB-D reference view and without any CAD models. It introduces the first end-to-end paradigm that operates entirely without CAD-based rendering, aligning reference and query observations through projective equivariant spatial alignment. The method effectively addresses challenges posed by symmetry, textureless surfaces, and occlusion by integrating three levels of semantic priors: category- or scene-level, symmetry-aware, and patch-level. Experimental results demonstrate that OneViewAll achieves a 92.5% ADD-0.1 accuracy on LINEMOD using a single reference view, substantially outperforming the baseline One2Any (52.6%), and consistently improves performance across multiple datasets—including YCB-V, Real275, and Toyota-Light—while maintaining low inference latency.

Abstract

In many practical 6D object pose estimation scenarios, we often have access to only a single real-world RGB-D reference view per object, typically without CAD models. Existing methods largely rely on explicit 3D models or multi-view data, which limits their scalability. To address this challenging single-reference model-free setting, we propose \textbf{OneViewAll}, a semantic-prior-guided framework that performs pose estimation via a novel Project-and-Compare paradigm. Instead of relying on computationally expensive CAD-based rendering, our method directly aligns reference and query observations within a projection-equivariant space. OneViewAll progressively integrates hierarchical semantic priors across three levels: (1) \textit{category- and scene-level} priors for efficient hypothesis initialization; (2) \textit{object-level symmetry} priors for geometry completion via mirror fusion; and (3) \textit{patch-level} priors for discriminative refinement. Extensive experiments demonstrate that OneViewAll achieves \textbf{92.5\%} ADD-0.1 accuracy on the LINEMOD dataset using only one real reference view -- significantly outperforming the CVPR 2025 baseline One2Any (52.6\%). It also yields consistent improvements on YCB-V, Real275, and Toyota-Light while maintaining low inference latency. Our results underscore the efficacy of symmetry-aware projection in handling symmetric, texture-less, and occluded objects.

Problem

• 6D pose estimation
• single-view
• model-free
• novel objects
• RGB-D

Innovation

• 6D pose estimation
• semantic prior
• single-view
• projection-equivariant
• symmetry-aware