OVE6D: Object Viewpoint Encoding for Depth-based 6D Object Pose Estimation

TL;DR

OVE6D is a lightweight, synthetic-data-trained framework that accurately estimates 6D object poses from depth images, generalizing well to new objects without real-world fine-tuning.

Contribution

It introduces a novel viewpoint encoding approach with lightweight modules, enabling robust, dataset-agnostic 6D pose estimation from synthetic data.

Findings

Outperforms some methods trained on real data

Generalizes well to unseen objects

Operates with less than 4 million parameters

Abstract

This paper proposes a universal framework, called OVE6D, for model-based 6D object pose estimation from a single depth image and a target object mask. Our model is trained using purely synthetic data rendered from ShapeNet, and, unlike most of the existing methods, it generalizes well on new real-world objects without any fine-tuning. We achieve this by decomposing the 6D pose into viewpoint, in-plane rotation around the camera optical axis and translation, and introducing novel lightweight modules for estimating each component in a cascaded manner. The resulting network contains less than 4M parameters while demonstrating excellent performance on the challenging T-LESS and Occluded LINEMOD datasets without any dataset-specific training. We show that OVE6D outperforms some contemporary deep learning-based pose estimation methods specifically trained for individual objects or datasets with real-world training data. The implementation and the pre-trained model will be made publicly available.