Generative 6d Pose Estimation via Conditional Flow Matching
MCML Authors
Abstract
Abstract
Existing methods for instance-level 6D pose estimation typically rely on neural networks that either directly regress the pose in SE(3) or estimate it indirectly via local feature matching. The former struggle with object symmetries, while the latter fail in the absence of distinctive local features. To overcome these limitations, we propose a novel formulation of 6D pose estimation as a conditional flow matching problem in ℝ3. We introduce Flose, a generative method that infers object poses through a denoising process conditioned on both geometric and appearance-based semantic features. Unlike prior conditional flow matching approaches, which rely solely on geometric guidance, Flose leverages semantic features to mitigate ambiguities caused by object symmetries while incorporating RANSAC-based registration for robustness against outliers. We validate Flose on five datasets from the established BOP benchmark, outperforming prior methods with an average improvement of +4.5 in Average Recall.
inproceedings HBL+26
ICIP 2026
IEEE International Conference on Image Processing. Tampere, Finland, Sep 13-17, 2026.Authors
A. Hamza • D. Boscaini • W. Li • B. Busam • F. PoiesiLinks
DOI GitHubResearch Area
BibTeXKey: HBL+26