TGNet: Learning 3D Shape from Sparse and Incomplete Point Cloud

Yushi Li*, Jia Wang, Yunzhe Wang, Rong Xiang, Yihong Wang, Yushan Pan

*Corresponding author for this work

Research output: Chapter in Book or Report/Conference proceedingConference Proceedingpeer-review

2 Citations (Scopus)

Abstract

Learning accurate 3D shapes from sparse and incomplete point clouds is challenging and meaningful, on account that the point clouds with low resolution always lack representative and informative details. This paper presents a novel deep auto-encoder called TGNet, which is formulated based on a tree-based generative adversarial network (GAN), to address self-supervised learning tasks on the point cloud with low sparsity. On the encoder side, we employ a PointNet-based framework to intensively capture the global representations. To better infer the spatial information in latent space, we propose a spectral graph learning module in with due consideration to graph topology. Further, we present a new loss that combines Wasserstein metric and multi-resolution Chamfer distance to better estimate global 3D geometry and structural details. The proposed TGNet achieves state-of-the-art performance for various point cloud learning tasks. Qualitative and quantitative evaluations demonstrate the novelty of the proposed model.

Original languageEnglish
Title of host publication2023 9th International Conference on Virtual Reality, ICVR 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages64-70
Number of pages7
ISBN (Electronic)9798350345810
DOIs
Publication statusPublished - 2023
Event9th International Conference on Virtual Reality, ICVR 2023 - Xianyang, China
Duration: 12 May 202314 May 2023

Publication series

Name2023 9th International Conference on Virtual Reality, ICVR 2023

Conference

Conference9th International Conference on Virtual Reality, ICVR 2023
Country/TerritoryChina
CityXianyang
Period12/05/2314/05/23

Keywords

  • Graph learning
  • Point cloud completion
  • Sparse Point Cloud Reconstruction

Fingerprint

Dive into the research topics of 'TGNet: Learning 3D Shape from Sparse and Incomplete Point Cloud'. Together they form a unique fingerprint.

Cite this