AlignPose: Generalizable 6D Pose Estimation via Multi-view Feature-metric Alignment

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mikeštíková, Anna Šárová, Fourmy, Médéric, Cífka, Martin, Sivic, Josef, Petrik, Vladimir
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910244169318400
author Mikeštíková, Anna Šárová
Fourmy, Médéric
Cífka, Martin
Sivic, Josef
Petrik, Vladimir
author_facet Mikeštíková, Anna Šárová
Fourmy, Médéric
Cífka, Martin
Sivic, Josef
Petrik, Vladimir
contents Single-view RGB model-based object pose estimation methods achieve strong generalization but are fundamentally limited by depth ambiguity, clutter, and occlusions. Multi-view pose estimation methods have the potential to solve these issues, but existing works rely on precise single-view pose estimates or lack generalization to unseen objects. We address these challenges via the following three contributions. First, we introduce AlignPose, a 6D object pose estimation method that aggregates information from multiple extrinsically calibrated RGB views and does not require any object-specific training or symmetry annotation. Second, the key component of this approach is a new multi-view feature-metric refinement specifically designed for object pose. It optimizes a single, consistent world-frame object pose by minimizing the feature discrepancy between on-the-fly rendered object features and observed image features across all views simultaneously. Third, we report extensive experiments on six datasets (YCB-V, T-LESS, HouseCat6D, ITODD-MV, IPD, XYZ-IBD) using the BOP benchmark evaluation and show that AlignPose outperforms other published methods, especially on challenging industrial datasets where multiple views are readily available in practice.
format Preprint
id arxiv_https___arxiv_org_abs_2512_20538
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AlignPose: Generalizable 6D Pose Estimation via Multi-view Feature-metric Alignment
Mikeštíková, Anna Šárová
Fourmy, Médéric
Cífka, Martin
Sivic, Josef
Petrik, Vladimir
Computer Vision and Pattern Recognition
Single-view RGB model-based object pose estimation methods achieve strong generalization but are fundamentally limited by depth ambiguity, clutter, and occlusions. Multi-view pose estimation methods have the potential to solve these issues, but existing works rely on precise single-view pose estimates or lack generalization to unseen objects. We address these challenges via the following three contributions. First, we introduce AlignPose, a 6D object pose estimation method that aggregates information from multiple extrinsically calibrated RGB views and does not require any object-specific training or symmetry annotation. Second, the key component of this approach is a new multi-view feature-metric refinement specifically designed for object pose. It optimizes a single, consistent world-frame object pose by minimizing the feature discrepancy between on-the-fly rendered object features and observed image features across all views simultaneously. Third, we report extensive experiments on six datasets (YCB-V, T-LESS, HouseCat6D, ITODD-MV, IPD, XYZ-IBD) using the BOP benchmark evaluation and show that AlignPose outperforms other published methods, especially on challenging industrial datasets where multiple views are readily available in practice.
title AlignPose: Generalizable 6D Pose Estimation via Multi-view Feature-metric Alignment
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.20538