Variance-preserving Deep Metric Learning for Content-based Image Retrieval

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

Standard

Variance-preserving Deep Metric Learning for Content-based Image Retrieval. / Passalis, Nikolaos; Iosifidis, Alexandros; Gabbouj, Moncef; Tefas, Anastasios.

I: Pattern Recognition Letters, Bind 131, 01.03.2020, s. 8-14.

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

Harvard

Passalis, N, Iosifidis, A, Gabbouj, M & Tefas, A 2020, 'Variance-preserving Deep Metric Learning for Content-based Image Retrieval', Pattern Recognition Letters, bind 131, s. 8-14. https://doi.org/10.1016/j.patrec.2019.11.041

APA

Passalis, N., Iosifidis, A., Gabbouj, M., & Tefas, A. (2020). Variance-preserving Deep Metric Learning for Content-based Image Retrieval. Pattern Recognition Letters, 131, 8-14. https://doi.org/10.1016/j.patrec.2019.11.041

CBE

MLA

Vancouver

Author

Passalis, Nikolaos ; Iosifidis, Alexandros ; Gabbouj, Moncef ; Tefas, Anastasios. / Variance-preserving Deep Metric Learning for Content-based Image Retrieval. I: Pattern Recognition Letters. 2020 ; Bind 131. s. 8-14.

Bibtex

@article{22c17c95e3ea4780ab146f3c41d09c2b,
title = "Variance-preserving Deep Metric Learning for Content-based Image Retrieval",
abstract = "Supervised deep metric learning led to spectacular results for several Content-based Information Retrieval (CBIR) applications. The success of these approaches slowly led to the belief that image retrieval and classification are just slightly different variations of the same problem. However, recent evidence suggests that learning highly discriminative representation for a (limited) set of training classes removes valuable information from the representation, potentially harming both the in-domain, as well as the out-of-domain retrieval precision. In this paper, we propose a regularized discriminative deep metric learning method that aims to not only learn a representation that allows for discriminating between different classes, but it is also capable of encoding the latent generative factors separately for each class, overcoming this limitation. This allows for modeling the in-class variance and, as a result, maintaining the ability to represent both sub-classes of the in-domain data, as well as objects that belong to classes outside the training domain. The effectiveness of the proposed method, over existing supervised and unsupervised representation/metric learning approaches, is demonstrated under different in-domain and out-of-domain setups and three challenging image datasets.",
keywords = "Content-based information retrieval, Deep learning, Metric learning",
author = "Nikolaos Passalis and Alexandros Iosifidis and Moncef Gabbouj and Anastasios Tefas",
year = "2020",
month = mar,
day = "1",
doi = "10.1016/j.patrec.2019.11.041",
language = "English",
volume = "131",
pages = "8--14",
journal = "Pattern Recognition Letters",
issn = "0167-8655",
publisher = "Elsevier BV * North-Holland",

}

RIS

TY - JOUR

T1 - Variance-preserving Deep Metric Learning for Content-based Image Retrieval

AU - Passalis, Nikolaos

AU - Iosifidis, Alexandros

AU - Gabbouj, Moncef

AU - Tefas, Anastasios

PY - 2020/3/1

Y1 - 2020/3/1

N2 - Supervised deep metric learning led to spectacular results for several Content-based Information Retrieval (CBIR) applications. The success of these approaches slowly led to the belief that image retrieval and classification are just slightly different variations of the same problem. However, recent evidence suggests that learning highly discriminative representation for a (limited) set of training classes removes valuable information from the representation, potentially harming both the in-domain, as well as the out-of-domain retrieval precision. In this paper, we propose a regularized discriminative deep metric learning method that aims to not only learn a representation that allows for discriminating between different classes, but it is also capable of encoding the latent generative factors separately for each class, overcoming this limitation. This allows for modeling the in-class variance and, as a result, maintaining the ability to represent both sub-classes of the in-domain data, as well as objects that belong to classes outside the training domain. The effectiveness of the proposed method, over existing supervised and unsupervised representation/metric learning approaches, is demonstrated under different in-domain and out-of-domain setups and three challenging image datasets.

AB - Supervised deep metric learning led to spectacular results for several Content-based Information Retrieval (CBIR) applications. The success of these approaches slowly led to the belief that image retrieval and classification are just slightly different variations of the same problem. However, recent evidence suggests that learning highly discriminative representation for a (limited) set of training classes removes valuable information from the representation, potentially harming both the in-domain, as well as the out-of-domain retrieval precision. In this paper, we propose a regularized discriminative deep metric learning method that aims to not only learn a representation that allows for discriminating between different classes, but it is also capable of encoding the latent generative factors separately for each class, overcoming this limitation. This allows for modeling the in-class variance and, as a result, maintaining the ability to represent both sub-classes of the in-domain data, as well as objects that belong to classes outside the training domain. The effectiveness of the proposed method, over existing supervised and unsupervised representation/metric learning approaches, is demonstrated under different in-domain and out-of-domain setups and three challenging image datasets.

KW - Content-based information retrieval

KW - Deep learning

KW - Metric learning

U2 - 10.1016/j.patrec.2019.11.041

DO - 10.1016/j.patrec.2019.11.041

M3 - Journal article

VL - 131

SP - 8

EP - 14

JO - Pattern Recognition Letters

JF - Pattern Recognition Letters

SN - 0167-8655

ER -