Object Level Depth Reconstruction for Category Level 6D Object Pose Estimation From Monocular RGB Image

TL;DR

This paper introduces OLD-Net, a novel RGB-only approach for category-level 6D object pose estimation that predicts object-level depth directly from monocular images, eliminating the need for depth sensors.

Contribution

The paper proposes a new RGB-based method with two innovative modules for high-fidelity depth and shape reconstruction, advancing category-level 6D pose estimation.

Findings

Achieves state-of-the-art results on CAMERA25 and REAL275 datasets.

Effectively predicts object-level depth and shape from monocular RGB images.

Outperforms existing RGBD-based methods despite using only RGB input.

Abstract

Recently, RGBD-based category-level 6D object pose estimation has achieved promising improvement in performance, however, the requirement of depth information prohibits broader applications. In order to relieve this problem, this paper proposes a novel approach named Object Level Depth reconstruction Network (OLD-Net) taking only RGB images as input for category-level 6D object pose estimation. We propose to directly predict object-level depth from a monocular RGB image by deforming the category-level shape prior into object-level depth and the canonical NOCS representation. Two novel modules named Normalized Global Position Hints (NGPH) and Shape-aware Decoupled Depth Reconstruction (SDDR) module are introduced to learn high fidelity object-level depth and delicate shape representations. At last, the 6D object pose is solved by aligning the predicted canonical representation with the back-projected object-level depth. Extensive experiments on the challenging CAMERA25 and REAL275 datasets indicate that our model, though simple, achieves state-of-the-art performance.