Dilated convolutions for image classification and object localization

Yasunori Kudo; Yoshimitsu Aoki

doi:10.23919/MVA.2017.7986898

Dilated convolutions for image classification and object localization

Yasunori Kudo, Yoshimitsu Aoki

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

15 Citations (Scopus)

Abstract

Yu et al.[1] showed that dilated convolutions are very effective in dense prediction problems such as semantic segmentation. In this work, we propose a new ResNet[2] based convolutional neural network model using dilated convolutions and show that this model can achieve lower error rate for image classification than ResNet with reduction of the number of the parameters of the network by 94% and that this model has high ability to localize objects despite being trained on image-level labels. We evaluated this model on ImageNet[5] which has 50 class labels randomly selected from 1000 class labels.

Original language	English
Title of host publication	Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	452-455
Number of pages	4
ISBN (Electronic)	9784901122160
DOIs	https://doi.org/10.23919/MVA.2017.7986898
Publication status	Published - 2017 Jul 19
Event	15th IAPR International Conference on Machine Vision Applications, MVA 2017 - Nagoya, Japan Duration: 2017 May 8 → 2017 May 12

Other

Other	15th IAPR International Conference on Machine Vision Applications, MVA 2017
Country/Territory	Japan
City	Nagoya
Period	17/5/8 → 17/5/12

ASJC Scopus subject areas

Computer Science Applications
Computer Vision and Pattern Recognition

Access to Document

10.23919/MVA.2017.7986898

Cite this

Dilated convolutions for image classification and object localization. / Kudo, Yasunori; Aoki, Yoshimitsu.

Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017. Institute of Electrical and Electronics Engineers Inc., 2017. p. 452-455 7986898.

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

Kudo, Y & Aoki, Y 2017, Dilated convolutions for image classification and object localization. in Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017., 7986898, Institute of Electrical and Electronics Engineers Inc., pp. 452-455, 15th IAPR International Conference on Machine Vision Applications, MVA 2017, Nagoya, Japan, 17/5/8. https://doi.org/10.23919/MVA.2017.7986898

@inproceedings{79bb132754234f1dbf7141696f688ff5,

title = "Dilated convolutions for image classification and object localization",

abstract = "Yu et al.[1] showed that dilated convolutions are very effective in dense prediction problems such as semantic segmentation. In this work, we propose a new ResNet[2] based convolutional neural network model using dilated convolutions and show that this model can achieve lower error rate for image classification than ResNet with reduction of the number of the parameters of the network by 94% and that this model has high ability to localize objects despite being trained on image-level labels. We evaluated this model on ImageNet[5] which has 50 class labels randomly selected from 1000 class labels.",

author = "Yasunori Kudo and Yoshimitsu Aoki",

year = "2017",

month = jul,

day = "19",

doi = "10.23919/MVA.2017.7986898",

language = "English",

pages = "452--455",

booktitle = "Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017",

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

note = "15th IAPR International Conference on Machine Vision Applications, MVA 2017 ; Conference date: 08-05-2017 Through 12-05-2017",

}

TY - GEN

T1 - Dilated convolutions for image classification and object localization

AU - Kudo, Yasunori

AU - Aoki, Yoshimitsu

PY - 2017/7/19

Y1 - 2017/7/19

N2 - Yu et al.[1] showed that dilated convolutions are very effective in dense prediction problems such as semantic segmentation. In this work, we propose a new ResNet[2] based convolutional neural network model using dilated convolutions and show that this model can achieve lower error rate for image classification than ResNet with reduction of the number of the parameters of the network by 94% and that this model has high ability to localize objects despite being trained on image-level labels. We evaluated this model on ImageNet[5] which has 50 class labels randomly selected from 1000 class labels.

AB - Yu et al.[1] showed that dilated convolutions are very effective in dense prediction problems such as semantic segmentation. In this work, we propose a new ResNet[2] based convolutional neural network model using dilated convolutions and show that this model can achieve lower error rate for image classification than ResNet with reduction of the number of the parameters of the network by 94% and that this model has high ability to localize objects despite being trained on image-level labels. We evaluated this model on ImageNet[5] which has 50 class labels randomly selected from 1000 class labels.

UR - http://www.scopus.com/inward/record.url?scp=85027884732&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027884732&partnerID=8YFLogxK

U2 - 10.23919/MVA.2017.7986898

DO - 10.23919/MVA.2017.7986898

M3 - Conference contribution

AN - SCOPUS:85027884732

SP - 452

EP - 455

BT - Proceedings of the 15th IAPR International Conference on Machine Vision Applications, MVA 2017

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 15th IAPR International Conference on Machine Vision Applications, MVA 2017

Y2 - 8 May 2017 through 12 May 2017

ER -

Dilated convolutions for image classification and object localization

Abstract

Other

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this