Relative attributes with deep Convolutional Neural Network

Dong Jin Kim; Donggeun Yoo; Sunghoon Im; Namil Kim; Tharatch Sirinukulwattana; In So Kweon

doi:10.1109/URAI.2015.7358851

Relative attributes with deep Convolutional Neural Network

Dong Jin Kim, Donggeun Yoo, Sunghoon Im, Namil Kim, Tharatch Sirinukulwattana, In So Kweon

Department of Electrical Engineering and Computer Science

Korea Advanced Institute of Science and Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Scopus citations

Abstract

Our work is based on the idea of relative attributes, aiming to provide more descriptive information to the images. We propose the model that integrates relative-attribute framework with deep Convolutional Neural Networks (CNN) to increase the accuracy of attribute comparison. In addition, we analyzed the role of each network layer in the process. Our model uses features extracted from CNN and is learned by Rank SVM method with these feature vectors. As a result, our model outperforms the original relative attribute model in terms of significant improvement in accuracy.

Original language	English
Title of host publication	2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	157-158
Number of pages	2
ISBN (Electronic)	9781467379700
DOIs	https://doi.org/10.1109/URAI.2015.7358851
State	Published - 16 Dec 2015
Event	12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015 - Goyang City, Korea, Republic of Duration: 28 Oct 2015 → 30 Oct 2015

Publication series

Name	2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015

Conference

Conference	12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015
Country/Territory	Korea, Republic of
City	Goyang City
Period	28/10/15 → 30/10/15

Bibliographical note

Publisher Copyright:
© 2015 IEEE.

Keywords

Deep learning
convolutional neural networks
relative attributes

Access to Document

10.1109/URAI.2015.7358851

Cite this

Kim, D. J., Yoo, D., Im, S., Kim, N., Sirinukulwattana, T., & Kweon, I. S. (2015). Relative attributes with deep Convolutional Neural Network. In 2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015 (pp. 157-158). Article 7358851 (2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/URAI.2015.7358851

Kim, Dong Jin ; Yoo, Donggeun ; Im, Sunghoon et al. / Relative attributes with deep Convolutional Neural Network. 2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015. Institute of Electrical and Electronics Engineers Inc., 2015. pp. 157-158 (2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015).

@inproceedings{9d5b6cc5f552476cb82639b69177b1ee,

title = "Relative attributes with deep Convolutional Neural Network",

abstract = "Our work is based on the idea of relative attributes, aiming to provide more descriptive information to the images. We propose the model that integrates relative-attribute framework with deep Convolutional Neural Networks (CNN) to increase the accuracy of attribute comparison. In addition, we analyzed the role of each network layer in the process. Our model uses features extracted from CNN and is learned by Rank SVM method with these feature vectors. As a result, our model outperforms the original relative attribute model in terms of significant improvement in accuracy.",

keywords = "Deep learning, convolutional neural networks, relative attributes",

author = "Kim, {Dong Jin} and Donggeun Yoo and Sunghoon Im and Namil Kim and Tharatch Sirinukulwattana and Kweon, {In So}",

note = "Publisher Copyright: {\textcopyright} 2015 IEEE.; 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015 ; Conference date: 28-10-2015 Through 30-10-2015",

year = "2015",

month = dec,

day = "16",

doi = "10.1109/URAI.2015.7358851",

language = "English",

series = "2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "157--158",

booktitle = "2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015",

}

Kim, DJ, Yoo, D, Im, S, Kim, N, Sirinukulwattana, T & Kweon, IS 2015, Relative attributes with deep Convolutional Neural Network. in 2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015., 7358851, 2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015, Institute of Electrical and Electronics Engineers Inc., pp. 157-158, 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015, Goyang City, Korea, Republic of, 28/10/15. https://doi.org/10.1109/URAI.2015.7358851

Relative attributes with deep Convolutional Neural Network. / Kim, Dong Jin; Yoo, Donggeun; Im, Sunghoon et al.
2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015. Institute of Electrical and Electronics Engineers Inc., 2015. p. 157-158 7358851 (2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Relative attributes with deep Convolutional Neural Network

AU - Kim, Dong Jin

AU - Yoo, Donggeun

AU - Im, Sunghoon

AU - Kim, Namil

AU - Sirinukulwattana, Tharatch

AU - Kweon, In So

PY - 2015/12/16

Y1 - 2015/12/16

N2 - Our work is based on the idea of relative attributes, aiming to provide more descriptive information to the images. We propose the model that integrates relative-attribute framework with deep Convolutional Neural Networks (CNN) to increase the accuracy of attribute comparison. In addition, we analyzed the role of each network layer in the process. Our model uses features extracted from CNN and is learned by Rank SVM method with these feature vectors. As a result, our model outperforms the original relative attribute model in terms of significant improvement in accuracy.

AB - Our work is based on the idea of relative attributes, aiming to provide more descriptive information to the images. We propose the model that integrates relative-attribute framework with deep Convolutional Neural Networks (CNN) to increase the accuracy of attribute comparison. In addition, we analyzed the role of each network layer in the process. Our model uses features extracted from CNN and is learned by Rank SVM method with these feature vectors. As a result, our model outperforms the original relative attribute model in terms of significant improvement in accuracy.

KW - Deep learning

KW - convolutional neural networks

KW - relative attributes

UR - http://www.scopus.com/inward/record.url?scp=84962696286&partnerID=8YFLogxK

U2 - 10.1109/URAI.2015.7358851

DO - 10.1109/URAI.2015.7358851

M3 - Conference contribution

AN - SCOPUS:84962696286

T3 - 2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015

SP - 157

EP - 158

BT - 2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015

Y2 - 28 October 2015 through 30 October 2015

ER -

Kim DJ, Yoo D, Im S, Kim N, Sirinukulwattana T, Kweon IS. Relative attributes with deep Convolutional Neural Network. In 2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015. Institute of Electrical and Electronics Engineers Inc. 2015. p. 157-158. 7358851. (2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2015). doi: 10.1109/URAI.2015.7358851

Relative attributes with deep Convolutional Neural Network

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this