End-to-End Probabilistic Geometry-Guided Regression for 6DoF Object Pose Estimation

TL;DR

This paper introduces EPRO-GDR, an end-to-end probabilistic approach for 6D object pose estimation that predicts a distribution of poses, improving accuracy and providing multiple plausible pose estimates.

Contribution

The paper presents a novel probabilistic regression method that estimates pose distributions instead of single poses, enhancing 6D pose estimation accuracy and robustness.

Findings

Superior quantitative results on multiple datasets

Ability to sample multiple plausible poses

Improved accuracy over state-of-the-art methods

Abstract

6D object pose estimation is the problem of identifying the position and orientation of an object relative to a chosen coordinate system, which is a core technology for modern XR applications. State-of-the-art 6D object pose estimators directly predict an object pose given an object observation. Due to the ill-posed nature of the pose estimation problem, where multiple different poses can correspond to a single observation, generating additional plausible estimates per observation can be valuable. To address this, we reformulate the state-of-the-art algorithm GDRNPP and introduce EPRO-GDR (End-to-End Probabilistic Geometry-Guided Regression). Instead of predicting a single pose per detection, we estimate a probability density distribution of the pose. Using the evaluation procedure defined by the BOP (Benchmark for 6D Object Pose Estimation) Challenge, we test our approach on four of its core datasets and demonstrate superior quantitative results for EPRO-GDR on LM-O, YCB-V, and ITODD. Our probabilistic solution shows that predicting a pose distribution instead of a single pose can improve state-of-the-art single-view pose estimation while providing the additional benefit of being able to sample multiple meaningful pose candidates.