PPConv: Polypod convolution for 3D point cloud description

Hyunsoo Song; Seungkyu Lee

doi:10.1145/3283289.3283362

PPConv: Polypod convolution for 3D point cloud description

Hyunsoo Song
, Seungkyu Lee

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

3D point cloud is a collection of unordered sparse 3D points that is different from densely structured color image. Therefore, applying a fixed structure of deep learning network on 3D point cloud is a challenging task in computer vision and graphics problems. Recently, researchers have proposed deep learning methods for 3D point cloud based on data conversion or simplification. However, they lose either local 3D shape information for the simplicity of method or geometric locality for using array as an input. In this paper we propose a new convolution technique, named Polypod convolution, for 3D point cloud description that is distribution independent and maintains both local and global 3D shapes. Quantitative and qualitative evaluation results show the potential of our new network for 3D point cloud based deep learning applications.

Original language	English
Title of host publication	SIGGRAPH Asia 2018 Posters, SA 2018
Publisher	Association for Computing Machinery, Inc
ISBN (Electronic)	9781450360630
DOIs	https://doi.org/10.1145/3283289.3283362
Publication status	Published - 4 Dec 2018
Event	SIGGRAPH Asia 2018 Posters - International Conference on Computer Graphics and Interactive Techniques, SA 2018 - Tokyo, Japan Duration: 4 Dec 2018 → 7 Dec 2018

Publication series

Name	SIGGRAPH Asia 2018 Posters, SA 2018

Conference

Conference	SIGGRAPH Asia 2018 Posters - International Conference on Computer Graphics and Interactive Techniques, SA 2018
Country/Territory	Japan
City	Tokyo
Period	4/12/18 → 7/12/18

Bibliographical note

Publisher Copyright:
© 2018 Copyright held by the owner/author(s).

Keywords

3d geometric
3d shape
Neural network
Point cloud

Access to Document

10.1145/3283289.3283362

Cite this

@inproceedings{e89a530b7e954490ba2fae81a93524de,

title = "PPConv: Polypod convolution for 3D point cloud description",

abstract = "3D point cloud is a collection of unordered sparse 3D points that is different from densely structured color image. Therefore, applying a fixed structure of deep learning network on 3D point cloud is a challenging task in computer vision and graphics problems. Recently, researchers have proposed deep learning methods for 3D point cloud based on data conversion or simplification. However, they lose either local 3D shape information for the simplicity of method or geometric locality for using array as an input. In this paper we propose a new convolution technique, named Polypod convolution, for 3D point cloud description that is distribution independent and maintains both local and global 3D shapes. Quantitative and qualitative evaluation results show the potential of our new network for 3D point cloud based deep learning applications.",

keywords = "3d geometric, 3d shape, Neural network, Point cloud",

author = "Hyunsoo Song and Seungkyu Lee",

note = "Publisher Copyright: {\textcopyright} 2018 Copyright held by the owner/author(s).; SIGGRAPH Asia 2018 Posters - International Conference on Computer Graphics and Interactive Techniques, SA 2018 ; Conference date: 04-12-2018 Through 07-12-2018",

year = "2018",

month = dec,

day = "4",

doi = "10.1145/3283289.3283362",

language = "English",

series = "SIGGRAPH Asia 2018 Posters, SA 2018",

publisher = "Association for Computing Machinery, Inc",

booktitle = "SIGGRAPH Asia 2018 Posters, SA 2018",

}

Song, H & Lee, S 2018, PPConv: Polypod convolution for 3D point cloud description. in SIGGRAPH Asia 2018 Posters, SA 2018., 3283362, SIGGRAPH Asia 2018 Posters, SA 2018, Association for Computing Machinery, Inc, SIGGRAPH Asia 2018 Posters - International Conference on Computer Graphics and Interactive Techniques, SA 2018, Tokyo, Japan, 4/12/18. https://doi.org/10.1145/3283289.3283362

TY - GEN

T1 - PPConv

T2 - SIGGRAPH Asia 2018 Posters - International Conference on Computer Graphics and Interactive Techniques, SA 2018

AU - Song, Hyunsoo

AU - Lee, Seungkyu

PY - 2018/12/4

Y1 - 2018/12/4

N2 - 3D point cloud is a collection of unordered sparse 3D points that is different from densely structured color image. Therefore, applying a fixed structure of deep learning network on 3D point cloud is a challenging task in computer vision and graphics problems. Recently, researchers have proposed deep learning methods for 3D point cloud based on data conversion or simplification. However, they lose either local 3D shape information for the simplicity of method or geometric locality for using array as an input. In this paper we propose a new convolution technique, named Polypod convolution, for 3D point cloud description that is distribution independent and maintains both local and global 3D shapes. Quantitative and qualitative evaluation results show the potential of our new network for 3D point cloud based deep learning applications.

AB - 3D point cloud is a collection of unordered sparse 3D points that is different from densely structured color image. Therefore, applying a fixed structure of deep learning network on 3D point cloud is a challenging task in computer vision and graphics problems. Recently, researchers have proposed deep learning methods for 3D point cloud based on data conversion or simplification. However, they lose either local 3D shape information for the simplicity of method or geometric locality for using array as an input. In this paper we propose a new convolution technique, named Polypod convolution, for 3D point cloud description that is distribution independent and maintains both local and global 3D shapes. Quantitative and qualitative evaluation results show the potential of our new network for 3D point cloud based deep learning applications.

KW - 3d geometric

KW - 3d shape

KW - Neural network

KW - Point cloud

UR - https://www.scopus.com/pages/publications/85060174229

U2 - 10.1145/3283289.3283362

DO - 10.1145/3283289.3283362

M3 - Conference contribution

AN - SCOPUS:85060174229

T3 - SIGGRAPH Asia 2018 Posters, SA 2018

BT - SIGGRAPH Asia 2018 Posters, SA 2018

PB - Association for Computing Machinery, Inc

Y2 - 4 December 2018 through 7 December 2018

ER -

PPConv: Polypod convolution for 3D point cloud description

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this