On the ability of a CNN to realize image-to-image language conversion

Kohei Baba; Seiichi Uchida; Brian Kenji Iwana

doi:10.1109/ICDAR.2019.00078

On the ability of a CNN to realize image-to-image language conversion

Kohei Baba, Seiichi Uchida, Brian Kenji Iwana

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

Abstract

The purpose of this paper is to reveal the ability that Convolutional Neural Networks (CNN) have on the novel task of image-to-image language conversion. We propose a new network to tackle this task by converting images of Korean Hangul characters directly into images of the phonetic Latin character equivalent. The conversion rules between Hangul and the phonetic symbols are not explicitly provided. The results of the proposed network show that it is possible to perform image-to-image language conversion. Moreover, it shows that it can grasp the structural features of Hangul even from limited learning data. In addition, it introduces a new network to use when the input and output have significantly different features.

Original language	English
Title of host publication	Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019
Publisher	IEEE Computer Society
Pages	448-453
Number of pages	6
ISBN (Electronic)	9781728128610
DOIs	https://doi.org/10.1109/ICDAR.2019.00078
Publication status	Published - Sept 2019
Event	15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019 - Sydney, Australia Duration: Sept 20 2019 → Sept 25 2019

Publication series

Name	Proceedings of the International Conference on Document Analysis and Recognition, ICDAR
ISSN (Print)	1520-5363

Conference

Conference	15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019
Country/Territory	Australia
City	Sydney
Period	9/20/19 → 9/25/19

All Science Journal Classification (ASJC) codes

Computer Vision and Pattern Recognition

Access to Document

10.1109/ICDAR.2019.00078

Cite this

Baba, K., Uchida, S., & Iwana, B. K. (2019). On the ability of a CNN to realize image-to-image language conversion. In Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019 (pp. 448-453). Article 8978060 (Proceedings of the International Conference on Document Analysis and Recognition, ICDAR). IEEE Computer Society. https://doi.org/10.1109/ICDAR.2019.00078

On the ability of a CNN to realize image-to-image language conversion. / Baba, Kohei; Uchida, Seiichi ; Iwana, Brian Kenji.
Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019. IEEE Computer Society, 2019. p. 448-453 8978060 (Proceedings of the International Conference on Document Analysis and Recognition, ICDAR).

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

Baba, K, Uchida, S & Iwana, BK 2019, On the ability of a CNN to realize image-to-image language conversion. in Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019., 8978060, Proceedings of the International Conference on Document Analysis and Recognition, ICDAR, IEEE Computer Society, pp. 448-453, 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019, Sydney, Australia, 9/20/19. https://doi.org/10.1109/ICDAR.2019.00078

Baba K, Uchida S , Iwana BK. On the ability of a CNN to realize image-to-image language conversion. In Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019. IEEE Computer Society. 2019. p. 448-453. 8978060. (Proceedings of the International Conference on Document Analysis and Recognition, ICDAR). doi: 10.1109/ICDAR.2019.00078

@inproceedings{739b2a32d180464488dc9c70e3a87603,

title = "On the ability of a CNN to realize image-to-image language conversion",

abstract = "The purpose of this paper is to reveal the ability that Convolutional Neural Networks (CNN) have on the novel task of image-to-image language conversion. We propose a new network to tackle this task by converting images of Korean Hangul characters directly into images of the phonetic Latin character equivalent. The conversion rules between Hangul and the phonetic symbols are not explicitly provided. The results of the proposed network show that it is possible to perform image-to-image language conversion. Moreover, it shows that it can grasp the structural features of Hangul even from limited learning data. In addition, it introduces a new network to use when the input and output have significantly different features.",

author = "Kohei Baba and Seiichi Uchida and Iwana, {Brian Kenji}",

note = "Funding Information: This work was supported by JSPS KAKENHI Grant Number JP17H06100. Publisher Copyright: {\textcopyright} 2019 IEEE.; 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019 ; Conference date: 20-09-2019 Through 25-09-2019",

year = "2019",

month = sep,

doi = "10.1109/ICDAR.2019.00078",

language = "English",

series = "Proceedings of the International Conference on Document Analysis and Recognition, ICDAR",

publisher = "IEEE Computer Society",

pages = "448--453",

booktitle = "Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019",

address = "United States",

}

TY - GEN

T1 - On the ability of a CNN to realize image-to-image language conversion

AU - Baba, Kohei

AU - Uchida, Seiichi

AU - Iwana, Brian Kenji

PY - 2019/9

Y1 - 2019/9

N2 - The purpose of this paper is to reveal the ability that Convolutional Neural Networks (CNN) have on the novel task of image-to-image language conversion. We propose a new network to tackle this task by converting images of Korean Hangul characters directly into images of the phonetic Latin character equivalent. The conversion rules between Hangul and the phonetic symbols are not explicitly provided. The results of the proposed network show that it is possible to perform image-to-image language conversion. Moreover, it shows that it can grasp the structural features of Hangul even from limited learning data. In addition, it introduces a new network to use when the input and output have significantly different features.

AB - The purpose of this paper is to reveal the ability that Convolutional Neural Networks (CNN) have on the novel task of image-to-image language conversion. We propose a new network to tackle this task by converting images of Korean Hangul characters directly into images of the phonetic Latin character equivalent. The conversion rules between Hangul and the phonetic symbols are not explicitly provided. The results of the proposed network show that it is possible to perform image-to-image language conversion. Moreover, it shows that it can grasp the structural features of Hangul even from limited learning data. In addition, it introduces a new network to use when the input and output have significantly different features.

UR - http://www.scopus.com/inward/record.url?scp=85079889083&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85079889083&partnerID=8YFLogxK

U2 - 10.1109/ICDAR.2019.00078

DO - 10.1109/ICDAR.2019.00078

M3 - Conference contribution

AN - SCOPUS:85079889083

T3 - Proceedings of the International Conference on Document Analysis and Recognition, ICDAR

SP - 448

EP - 453

BT - Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019

PB - IEEE Computer Society

T2 - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019

Y2 - 20 September 2019 through 25 September 2019

ER -

On the ability of a CNN to realize image-to-image language conversion

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this