3DPointCaps++: Learning 3D Representations with Capsule Networks

We present 3DPointCaps++ for learning robust, flexible and generalizable 3D object representations without requiring heavy annotation efforts or supervision. Unlike conventional 3D generative models, our algorithm aims for building a structured latent space where certain factors of shape variations,...

Full description

Saved in:
Bibliographic Details
Published in:International journal of computer vision Vol. 130; no. 9; pp. 2321 - 2336
Main Authors: Zhao, Yongheng, Fang, Guangchi, Guo, Yulan, Guibas, Leonidas, Tombari, Federico, Birdal, Tolga
Format: Journal Article
Language:English
Published: New York Springer US 01.09.2022
Springer
Springer Nature B.V
Subjects:
Algorithms
Annotations
Artificial Intelligence
Computer Imaging
Computer Science
Image Processing and Computer Vision
Interpolation
Learning
Pattern Recognition
Pattern Recognition and Graphics
Representations
Robustness
Segmentation
Special Issue on 3D Computer Vision
Three dimensional models
Vision
Representation learning
3D Shapes
3D Reconstruction
Autoencoder
3D Point clouds
Capsule networks
Unsupervised learning
ISSN:0920-5691, 1573-1405
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We present 3DPointCaps++ for learning robust, flexible and generalizable 3D object representations without requiring heavy annotation efforts or supervision. Unlike conventional 3D generative models, our algorithm aims for building a structured latent space where certain factors of shape variations, such as object parts, can be disentangled into independent sub-spaces. Our novel decoder then acts on these individual latent sub-spaces (i.e. capsules) using deconvolution operators to reconstruct 3D points in a self-supervised manner. We further introduce a cluster loss ensuring that the points reconstructed by a single capsule remain local and do not spread across the object uncontrollably. These contributions allow our network to tackle the challenging tasks of part segmentation, part interpolation/replacement as well as correspondence estimation across rigid / non-rigid shape, and across / within category. Our extensive evaluations on ShapeNet objects and human scans demonstrate that our network can learn generic representations that are robust and useful in many applications.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
Communicated by Akihiro Sugimoto.
ISSN:0920-5691
1573-1405
DOI:10.1007/s11263-022-01632-6