Skip to main navigation Skip to search Skip to main content

Semi-supervised generative adversarial model for 3D recognition

  • Tsinghua University
  • Shenzhen Institute of Future Media Technology

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

Abstract

In this paper, we focus on the issue of 3D point cloud recognition utilizing semi-supervised generative adversarial network, composed of two parts: classification net and generative adversarial net. Using the adversarial criterion, the generator learns to establish the mapping relationship from one-dimensional noise data to 3D point cloud, which can remedy the insufficient data available faced by traditional supervised methods, and the adversarial discriminator acts as a powerful 3D object descriptor to get the point cloud into the right category and eliminate the fake data brought by the generator. In order to fix the rotation problem of point cloud data and improve the recognition performance of the model, we also introduce the spatial transformation networks into the discriminator. Compared with the current supervised and unsupervised methods, the impressive experimental results we show illustrate the effectiveness of the proposed method.

Original languageEnglish
Title of host publicationProceedings - 2019 IEEE International Conferences on Ubiquitous Computing and Communications and Data Science and Computational Intelligence and Smart Computing, Networking and Services, IUCC/DSCI/SmartCNS 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages381-385
Number of pages5
ISBN (Electronic)9781728152097
DOIs
StatePublished - Oct 2019
Externally publishedYes
Event2019 IEEE International Conferences on Ubiquitous Computing and Communications and Data Science and Computational Intelligence and Smart Computing, Networking and Services, IUCC/DSCI/SmartCNS 2019 - Shenyang, China
Duration: 21 Oct 201923 Oct 2019

Publication series

NameProceedings - 2019 IEEE International Conferences on Ubiquitous Computing and Communications and Data Science and Computational Intelligence and Smart Computing, Networking and Services, IUCC/DSCI/SmartCNS 2019

Conference

Conference2019 IEEE International Conferences on Ubiquitous Computing and Communications and Data Science and Computational Intelligence and Smart Computing, Networking and Services, IUCC/DSCI/SmartCNS 2019
Country/TerritoryChina
CityShenyang
Period21/10/1923/10/19

Keywords

  • 3D point cloud
  • Adversarial network
  • Classification
  • Generative
  • Semi-supervised
  • Spatial transformation network

Fingerprint

Dive into the research topics of 'Semi-supervised generative adversarial model for 3D recognition'. Together they form a unique fingerprint.

Cite this