Title:
SHAPE SUPPLEMENTATION DEVICE, SHAPE SUPPLEMENTATION LEARNING DEVICE, METHOD, AND PROGRAM
Document Type and Number:
WIPO Patent Application WO/2020/066662
Kind Code:
A1
Abstract:
An objective of the present invention is to enable precision shape supplementation with a point cloud as an input. A shape supplementation part: inputs an inputted point cloud and a class identification feature outputted by a class identification part into a generator which obtains, with a pre-learned point cloud and class identification feature as inputs, an integration result in which the class identification feature is integrated with a global feature which is based on local features extracted from each point of the point cloud and generates a shape supplementation point cloud which is a set of three-dimensional points for supplementing a point cloud by convolution of the integration result; and outputs the shape supplementation point cloud for supplementing the inputted point cloud.
Inventors:
NAGANO HIDEHISA (JP)
IRIE GO (JP)
ITO SEIYA (JP)
SUMI KAZUHIKO (JP)
IRIE GO (JP)
ITO SEIYA (JP)
SUMI KAZUHIKO (JP)
Application Number:
PCT/JP2019/035910
Publication Date:
April 02, 2020
Filing Date:
September 12, 2019
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G06T7/00; G06N20/00
Other References:
ITO, S. ET AL.: "Proposal of object recognition and shape completion network using point cloud as input", IPSJ SIG TECHNICAL REPORT, vol. 2018-CV, no. 54, 19 January 2018 (2018-01-19), pages 1 - 4, XP055700617, ISSN: 2188-8701, Retrieved from the Internet [retrieved on 20180117]
YUAN, W. ET AL.: "PCN: Point Completion Network", ARXIV:1808.00671V2, 2 August 2018 (2018-08-02), pages 1 - 17, XP081411641, Retrieved from the Internet [retrieved on 20191115]
KINGKAN, C. ET AL.: "Generating Mesh-based Shapes From Learned Latent Spaces of Point Clouds with VAE-GAN", PROCEEDINGS OF THE 2018 24TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR, 20 August 2018 (2018-08-20), pages 308 - 313, XP033459908, ISBN: 978-1-5386-3788-3, DOI: 10.1109/ICPR.2018.8546232
ACHLIOPTAS, P. ET AL.: "Learning Representations and Generative Models for 3D Point Clouds", ARXIV:1707.02392V3, 8 July 2018 (2018-07-08), pages 1 - 18, XP081324797, Retrieved from the Internet [retrieved on 20191115]
YANG, Y. ET AL.: "FoldingNet: Point Cloud Auto- encoder via Deep Grid Deformation", ARXIV:1712.07262V2, 3 April 2018 (2018-04-03), pages 1 - 14, XP081314317, Retrieved from the Internet [retrieved on 20191115]
QI, C. R. ET AL.: "PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation", ARXIV:1612.00593V2, 3 April 2018 (2018-04-03), pages 1 - 19, XP055700641, Retrieved from the Internet [retrieved on 20191115]
YUAN, W. ET AL.: "PCN: Point Completion Network", ARXIV:1808.00671V2, 2 August 2018 (2018-08-02), pages 1 - 17, XP081411641, Retrieved from the Internet
KINGKAN, C. ET AL.: "Generating Mesh-based Shapes From Learned Latent Spaces of Point Clouds with VAE-GAN", PROCEEDINGS OF THE 2018 24TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR, 20 August 2018 (2018-08-20), pages 308 - 313, XP033459908, ISBN: 978-1-5386-3788-3, DOI: 10.1109/ICPR.2018.8546232
ACHLIOPTAS, P. ET AL.: "Learning Representations and Generative Models for 3D Point Clouds", ARXIV:1707.02392V3, 8 July 2018 (2018-07-08), pages 1 - 18, XP081324797, Retrieved from the Internet
YANG, Y. ET AL.: "FoldingNet: Point Cloud Auto- encoder via Deep Grid Deformation", ARXIV:1712.07262V2, 3 April 2018 (2018-04-03), pages 1 - 14, XP081314317, Retrieved from the Internet
QI, C. R. ET AL.: "PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation", ARXIV:1612.00593V2, 3 April 2018 (2018-04-03), pages 1 - 19, XP055700641, Retrieved from the Internet
Attorney, Agent or Firm:
TAIYO, NAKAJIMA & KATO (JP)
Download PDF:
Previous Patent: LINE CARRYING SYSTEM
Next Patent: PLASMA ELECTRON BEAM TREATMENT INKJET-PRINTING DEVICE
Next Patent: PLASMA ELECTRON BEAM TREATMENT INKJET-PRINTING DEVICE