Primitive shape recognition via superquadric representation using large margin nearest neighbor classifier

Ryo Hachiuma, Yuko Ozasa, Hideo Saito

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

It is known that humans recognize objects using combinations and positional relations of primitive shapes. The first step of such recognition is to recognize 3D primitive shapes. In this paper, we propose a method for primitive shape recognition using superquadric parameters with a metric learning method, large margin nearest neighbor (LMNN). Superquadrics can represent various types of primitive shapes using a single equation with few parameters. These parameters are used as the feature vector of classification. The real objects of primitive shapes are used in our experiment, and the results show the effectiveness of using LMNN for recognition based on superquadrics. Compared to the previous methods, which used k-nearest neighbors (76.5%) and Support Vector Machines (73.5%), our LMNN method has the best performance (79.5%).

Original languageEnglish
Title of host publicationVISAPP
PublisherSciTePress
Pages325-332
Number of pages8
Volume5
ISBN (Electronic)9789897582264
Publication statusPublished - 2017 Jan 1
Event12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2017 - Porto, Portugal
Duration: 2017 Feb 272017 Mar 1

Other

Other12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2017
CountryPortugal
CityPorto
Period17/2/2717/3/1

Fingerprint

Support vector machines
Classifiers
Experiments

Keywords

  • 3D shape primitives
  • Large margin nearest neighbor
  • Primitive shape recognition
  • Superquadrics

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

Primitive shape recognition via superquadric representation using large margin nearest neighbor classifier. / Hachiuma, Ryo; Ozasa, Yuko; Saito, Hideo.

VISAPP. Vol. 5 SciTePress, 2017. p. 325-332.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hachiuma, R, Ozasa, Y & Saito, H 2017, Primitive shape recognition via superquadric representation using large margin nearest neighbor classifier. in VISAPP. vol. 5, SciTePress, pp. 325-332, 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2017, Porto, Portugal, 17/2/27.
@inproceedings{0da6b8b3f4ba4af6a7858e69035194de,
title = "Primitive shape recognition via superquadric representation using large margin nearest neighbor classifier",
abstract = "It is known that humans recognize objects using combinations and positional relations of primitive shapes. The first step of such recognition is to recognize 3D primitive shapes. In this paper, we propose a method for primitive shape recognition using superquadric parameters with a metric learning method, large margin nearest neighbor (LMNN). Superquadrics can represent various types of primitive shapes using a single equation with few parameters. These parameters are used as the feature vector of classification. The real objects of primitive shapes are used in our experiment, and the results show the effectiveness of using LMNN for recognition based on superquadrics. Compared to the previous methods, which used k-nearest neighbors (76.5{\%}) and Support Vector Machines (73.5{\%}), our LMNN method has the best performance (79.5{\%}).",
keywords = "3D shape primitives, Large margin nearest neighbor, Primitive shape recognition, Superquadrics",
author = "Ryo Hachiuma and Yuko Ozasa and Hideo Saito",
year = "2017",
month = "1",
day = "1",
language = "English",
volume = "5",
pages = "325--332",
booktitle = "VISAPP",
publisher = "SciTePress",

}

TY - GEN

T1 - Primitive shape recognition via superquadric representation using large margin nearest neighbor classifier

AU - Hachiuma, Ryo

AU - Ozasa, Yuko

AU - Saito, Hideo

PY - 2017/1/1

Y1 - 2017/1/1

N2 - It is known that humans recognize objects using combinations and positional relations of primitive shapes. The first step of such recognition is to recognize 3D primitive shapes. In this paper, we propose a method for primitive shape recognition using superquadric parameters with a metric learning method, large margin nearest neighbor (LMNN). Superquadrics can represent various types of primitive shapes using a single equation with few parameters. These parameters are used as the feature vector of classification. The real objects of primitive shapes are used in our experiment, and the results show the effectiveness of using LMNN for recognition based on superquadrics. Compared to the previous methods, which used k-nearest neighbors (76.5%) and Support Vector Machines (73.5%), our LMNN method has the best performance (79.5%).

AB - It is known that humans recognize objects using combinations and positional relations of primitive shapes. The first step of such recognition is to recognize 3D primitive shapes. In this paper, we propose a method for primitive shape recognition using superquadric parameters with a metric learning method, large margin nearest neighbor (LMNN). Superquadrics can represent various types of primitive shapes using a single equation with few parameters. These parameters are used as the feature vector of classification. The real objects of primitive shapes are used in our experiment, and the results show the effectiveness of using LMNN for recognition based on superquadrics. Compared to the previous methods, which used k-nearest neighbors (76.5%) and Support Vector Machines (73.5%), our LMNN method has the best performance (79.5%).

KW - 3D shape primitives

KW - Large margin nearest neighbor

KW - Primitive shape recognition

KW - Superquadrics

UR - http://www.scopus.com/inward/record.url?scp=85047869187&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85047869187&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85047869187

VL - 5

SP - 325

EP - 332

BT - VISAPP

PB - SciTePress

ER -