WAPR: A Foundation Model for Wide-Angle Refinement in Unseen Object Pose Estimation

2026-10-07

AI Summary

This study addresses the limited accuracy and efficiency of 6D pose estimation for unseen objects under large rotational deviations. We propose a zero-shot wide-angle pose refinement model that introduces a rotation symmetry prior to regularize the optimization objective and designs an angle-balanced loss to stabilize training, enabling correction of initial deviations up to 90°. Furthermore, we construct the SA6D dataset based on symmetry priors and enhance generalization by integrating geometric-texture augmentation with KASAL-assisted symmetry modeling. Extensive experiments demonstrate that the proposed method achieves state-of-the-art performance across seven BOP benchmarks while maintaining single-frame inference in under one second and a throughput of 25 instances per second, successfully unifying precision and computational efficiency.

Abstract

Real-world applications require 6D pose estimation to be accurate, fast, and scalable to unseen objects. This paper introduces WAPR, a zero-shot wide-angle pose refinement model that refines candidate poses with rotational deviations up to 90 degrees. With as few as 12 candidate poses per detected object instance, WAPR supports fast inference within 1 s per frame and reaches a pose-estimation throughput of up to 25 detected object instances per second. To support wide-angle training for rotationally symmetric objects, WAPR uses rotational symmetry priors to canonicalize symmetry-equivalent pose targets before loss computation. We further construct SA6D, a large-scale 6D training dataset with such priors. SA6D obtains KASAL-assisted rotational symmetry priors for 944 GSO scans and expands them through geometry and texture augmentation into about 50K augmented object instances and about 2M rendered RGB-D images. In addition, an angle-balanced loss stabilizes learning across different angular ranges by reducing the influence of uninformative large-error cases. Experiments on seven BOP core datasets show that WAPR achieves state-of-the-art performance in unseen-object 6D pose localization and detection under both fast and unconstrained inference settings. Project page: https://github.com/WangYuLin-SEU/WAPR.

Problem

• 6D pose estimation
• unseen objects
• pose refinement
• zero-shot
• rotational symmetry

Innovation

• Zero-shot Pose Refinement
• Wide-Angle Refinement
• Rotational Symmetry Priors
• Angle-Balanced Loss
• Large-scale 6D Dataset