DenseFusion: 6D Object Pose Estimation by Iterative Dense Fusion

TL;DR

DenseFusion introduces an iterative dense fusion framework that effectively combines RGB and depth data for accurate, real-time 6D object pose estimation, outperforming previous methods in cluttered scenes and practical robotic applications.

Contribution

The paper proposes a novel dense fusion network and iterative refinement process for improved 6D pose estimation from RGB-D images, enabling real-time performance.

Findings

Outperforms state-of-the-art on YCB-Video and LineMOD datasets

Achieves near real-time inference

Successfully applied to robotic grasping and manipulation

Abstract

A key technical challenge in performing 6D object pose estimation from RGB-D image is to fully leverage the two complementary data sources. Prior works either extract information from the RGB image and depth separately or use costly post-processing steps, limiting their performances in highly cluttered scenes and real-time applications. In this work, we present DenseFusion, a generic framework for estimating 6D pose of a set of known objects from RGB-D images. DenseFusion is a heterogeneous architecture that processes the two data sources individually and uses a novel dense fusion network to extract pixel-wise dense feature embedding, from which the pose is estimated. Furthermore, we integrate an end-to-end iterative pose refinement procedure that further improves the pose estimation while achieving near real-time inference. Our experiments show that our method outperforms state-of-the-art approaches in two datasets, YCB-Video and LineMOD. We also deploy our proposed method to a real robot to grasp and manipulate objects based on the estimated pose.