Deep Multimetric Learning for Shape-Based 3D Model Retrieval

Jin Xie, Guoxian Dai, Yi Fang

Research output: Contribution to journalArticle

Abstract

Recently, feature-learning-based 3D shape retrieval methods have been receiving more and more attention in the 3D shape analysis community. In these methods, the hand-crafted metrics or the learned linear metrics are usually used to compute the distances between shape features. Since there are complex geometric structural variations with 3D shapes, the single hand-crafted metric or learned linear metric cannot characterize the manifold, where 3D shapes lie well. In this paper, by exploring the nonlinearity of the deep neural network and the complementarity among multiple shape features, we propose a novel deep multimetric network for 3D shape retrieval. The developed multimetric network minimizes a discriminative loss function that, for each type of shape feature, the outputs of the network from the same class are encouraged to be as similar as possible and the outputs from different classes are encouraged to be as dissimilar as possible. Meanwhile, the Hilbert-Schmidt independence criterion is employed to enforce the outputs of different types of shape features to be as complementary as possible. Furthermore, the weights of the learned multiple distance metrics can be adaptively determined in our developed deep metric network. The weighted distance metric is then used as the similarity for shape retrieval. We conduct experiments with the proposed method on the four benchmark shape datasets. Experimental results demonstrate that the proposed method can obtain better performance than the learned deep single metric and outperform the state-of-the-art 3D shape retrieval methods.

Original languageEnglish (US)
Article number7912387
Pages (from-to)2463-2474
Number of pages12
JournalIEEE Transactions on Multimedia
Volume19
Issue number11
DOIs
StatePublished - Nov 1 2017

Fingerprint

Experiments
Deep learning
Deep neural networks

Keywords

  • 3D shape descriptor
  • 3D shape retrieval
  • deep neural network
  • metric learning
  • multiple shape features

ASJC Scopus subject areas

  • Signal Processing
  • Media Technology
  • Computer Science Applications
  • Electrical and Electronic Engineering

Cite this

Deep Multimetric Learning for Shape-Based 3D Model Retrieval. / Xie, Jin; Dai, Guoxian; Fang, Yi.

In: IEEE Transactions on Multimedia, Vol. 19, No. 11, 7912387, 01.11.2017, p. 2463-2474.

Research output: Contribution to journalArticle

Xie, Jin ; Dai, Guoxian ; Fang, Yi. / Deep Multimetric Learning for Shape-Based 3D Model Retrieval. In: IEEE Transactions on Multimedia. 2017 ; Vol. 19, No. 11. pp. 2463-2474.
@article{8e6393d922fe402fbe835d6c14f9165f,
title = "Deep Multimetric Learning for Shape-Based 3D Model Retrieval",
abstract = "Recently, feature-learning-based 3D shape retrieval methods have been receiving more and more attention in the 3D shape analysis community. In these methods, the hand-crafted metrics or the learned linear metrics are usually used to compute the distances between shape features. Since there are complex geometric structural variations with 3D shapes, the single hand-crafted metric or learned linear metric cannot characterize the manifold, where 3D shapes lie well. In this paper, by exploring the nonlinearity of the deep neural network and the complementarity among multiple shape features, we propose a novel deep multimetric network for 3D shape retrieval. The developed multimetric network minimizes a discriminative loss function that, for each type of shape feature, the outputs of the network from the same class are encouraged to be as similar as possible and the outputs from different classes are encouraged to be as dissimilar as possible. Meanwhile, the Hilbert-Schmidt independence criterion is employed to enforce the outputs of different types of shape features to be as complementary as possible. Furthermore, the weights of the learned multiple distance metrics can be adaptively determined in our developed deep metric network. The weighted distance metric is then used as the similarity for shape retrieval. We conduct experiments with the proposed method on the four benchmark shape datasets. Experimental results demonstrate that the proposed method can obtain better performance than the learned deep single metric and outperform the state-of-the-art 3D shape retrieval methods.",
keywords = "3D shape descriptor, 3D shape retrieval, deep neural network, metric learning, multiple shape features",
author = "Jin Xie and Guoxian Dai and Yi Fang",
year = "2017",
month = "11",
day = "1",
doi = "10.1109/TMM.2017.2698200",
language = "English (US)",
volume = "19",
pages = "2463--2474",
journal = "IEEE Transactions on Multimedia",
issn = "1520-9210",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "11",

}

TY - JOUR

T1 - Deep Multimetric Learning for Shape-Based 3D Model Retrieval

AU - Xie, Jin

AU - Dai, Guoxian

AU - Fang, Yi

PY - 2017/11/1

Y1 - 2017/11/1

N2 - Recently, feature-learning-based 3D shape retrieval methods have been receiving more and more attention in the 3D shape analysis community. In these methods, the hand-crafted metrics or the learned linear metrics are usually used to compute the distances between shape features. Since there are complex geometric structural variations with 3D shapes, the single hand-crafted metric or learned linear metric cannot characterize the manifold, where 3D shapes lie well. In this paper, by exploring the nonlinearity of the deep neural network and the complementarity among multiple shape features, we propose a novel deep multimetric network for 3D shape retrieval. The developed multimetric network minimizes a discriminative loss function that, for each type of shape feature, the outputs of the network from the same class are encouraged to be as similar as possible and the outputs from different classes are encouraged to be as dissimilar as possible. Meanwhile, the Hilbert-Schmidt independence criterion is employed to enforce the outputs of different types of shape features to be as complementary as possible. Furthermore, the weights of the learned multiple distance metrics can be adaptively determined in our developed deep metric network. The weighted distance metric is then used as the similarity for shape retrieval. We conduct experiments with the proposed method on the four benchmark shape datasets. Experimental results demonstrate that the proposed method can obtain better performance than the learned deep single metric and outperform the state-of-the-art 3D shape retrieval methods.

AB - Recently, feature-learning-based 3D shape retrieval methods have been receiving more and more attention in the 3D shape analysis community. In these methods, the hand-crafted metrics or the learned linear metrics are usually used to compute the distances between shape features. Since there are complex geometric structural variations with 3D shapes, the single hand-crafted metric or learned linear metric cannot characterize the manifold, where 3D shapes lie well. In this paper, by exploring the nonlinearity of the deep neural network and the complementarity among multiple shape features, we propose a novel deep multimetric network for 3D shape retrieval. The developed multimetric network minimizes a discriminative loss function that, for each type of shape feature, the outputs of the network from the same class are encouraged to be as similar as possible and the outputs from different classes are encouraged to be as dissimilar as possible. Meanwhile, the Hilbert-Schmidt independence criterion is employed to enforce the outputs of different types of shape features to be as complementary as possible. Furthermore, the weights of the learned multiple distance metrics can be adaptively determined in our developed deep metric network. The weighted distance metric is then used as the similarity for shape retrieval. We conduct experiments with the proposed method on the four benchmark shape datasets. Experimental results demonstrate that the proposed method can obtain better performance than the learned deep single metric and outperform the state-of-the-art 3D shape retrieval methods.

KW - 3D shape descriptor

KW - 3D shape retrieval

KW - deep neural network

KW - metric learning

KW - multiple shape features

UR - http://www.scopus.com/inward/record.url?scp=85032305262&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85032305262&partnerID=8YFLogxK

U2 - 10.1109/TMM.2017.2698200

DO - 10.1109/TMM.2017.2698200

M3 - Article

VL - 19

SP - 2463

EP - 2474

JO - IEEE Transactions on Multimedia

JF - IEEE Transactions on Multimedia

SN - 1520-9210

IS - 11

M1 - 7912387

ER -