Category-Level 6D Object Pose Estimation in the Wild: A Semi-Supervised Learning Approach and A New Dataset

TL;DR

This paper introduces Wild6D, a new unlabeled dataset for category-level 6D object pose estimation in diverse real-world scenarios, and proposes RePoNet, a semi-supervised model that outperforms existing methods without requiring 3D annotations on real data.

Contribution

The paper presents Wild6D dataset and a semi-supervised learning approach, RePoNet, for improved category-level 6D pose estimation in the wild.

Findings

RePoNet outperforms state-of-the-art methods on Wild6D and previous datasets.

Wild6D dataset contains diverse instances and backgrounds for robust training.

Semi-supervised learning reduces the need for extensive 3D annotations.

Abstract

6D object pose estimation is one of the fundamental problems in computer vision and robotics research. While a lot of recent efforts have been made on generalizing pose estimation to novel object instances within the same category, namely category-level 6D pose estimation, it is still restricted in constrained environments given the limited number of annotated data. In this paper, we collect Wild6D, a new unlabeled RGBD object video dataset with diverse instances and backgrounds. We utilize this data to generalize category-level 6D object pose estimation in the wild with semi-supervised learning. We propose a new model, called Rendering for Pose estimation network RePoNet, that is jointly trained using the free ground-truths with the synthetic data, and a silhouette matching objective function on the real-world data. Without using any 3D annotations on real data, our method outperforms state-of-the-art methods on the previous dataset and our Wild6D test set (with manual annotations for evaluation) by a large margin. Project page with Wild6D data: https://oasisyang.github.io/semi-pose .