LocPoseNet: Robust Location Prior for Unseen Object Pose Estimation

TL;DR

LocPoseNet is a novel approach that robustly predicts the location prior for unseen objects, significantly improving 6D pose estimation accuracy by leveraging a multi-scale correlation strategy and a decoupled translation estimator.

Contribution

It introduces a robust location prior learning method for unseen objects using template matching and a novel translation estimator, outperforming existing methods.

Findings

Outperforms existing methods on LINEMOD and GenMOP datasets.

Demonstrates robustness to various noise sources in synthetic datasets.

Provides a scalable and efficient multi-scale correlation computation.

Abstract

Object location prior is critical for the standard 6D object pose estimation setting. The prior can be used to initialize the 3D object translation and facilitate 3D object rotation estimation. Unfortunately, the object detectors that are used for this purpose do not generalize to unseen objects. Therefore, existing 6D pose estimation methods for unseen objects either assume the ground-truth object location to be known or yield inaccurate results when it is unavailable. In this paper, we address this problem by developing a method, LocPoseNet, able to robustly learn location prior for unseen objects. Our method builds upon a template matching strategy, where we propose to distribute the reference kernels and convolve them with a query to efficiently compute multi-scale correlations. We then introduce a novel translation estimator, which decouples scale-aware and scale-robust features to predict different object location parameters. Our method outperforms existing works by a large margin on LINEMOD and GenMOP. We further construct a challenging synthetic dataset, which allows us to highlight the better robustness of our method to various noise sources. Our project website is at: https://sailor-z.github.io/projects/3DV2024_LocPoseNet.html.