Classification of photo and sketch images using convolutional neural networks

Kazuma Sasaki; Madoka Yamakawa; Kana Sekiguchi; Tetsuya Ogata

doi:10.1007/978-3-319-44781-0_34

Classification of photo and sketch images using convolutional neural networks

Kazuma Sasaki, Madoka Yamakawa, Kana Sekiguchi, Tetsuya Ogata

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

2 Citations (Scopus)

Abstract

Content-Based Image Retrieval (CBIR) system enables us to access images using only images as queries, instead of keywords. Photorealistic images, and hand-drawn sketch image can be used as a queries as well. Recently, convolutional neural networks (CNNs) are used to discriminate images including sketches. However, the tasks are limited to classifying only one type of images, either photo or sketch images, due to the lack of a large dataset of sketch images and the large difference of their visual characteristics. In this paper, we introduce a simple way to prepare training datasets, which can enable the CNN model to classify both types of images by color transforming photo and illustration images. Through the training experiment, we show that the proposed method contributes to the improvement of classification accuracy.

Original language	English
Title of host publication	Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings
Publisher	Springer Verlag
Pages	283-290
Number of pages	8
Volume	9887 LNCS
ISBN (Print)	9783319447803
DOIs	https://doi.org/10.1007/978-3-319-44781-0_34
Publication status	Published - 2016
Event	25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016 - Barcelona, Spain Duration: 2016 Sep 6 → 2016 Sep 9

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9887 LNCS
ISSN (Print)	03029743
ISSN (Electronic)	16113349

Other

Other	25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016
Country	Spain
City	Barcelona
Period	16/9/6 → 16/9/9

Fingerprint

Keywords

Content based image retrieval
Hand-drawn sketch

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Cite this

Sasaki, K., Yamakawa, M., Sekiguchi, K., & Ogata, T. (2016). Classification of photo and sketch images using convolutional neural networks. In Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings (Vol. 9887 LNCS, pp. 283-290). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9887 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-44781-0_34

Classification of photo and sketch images using convolutional neural networks. / Sasaki, Kazuma; Yamakawa, Madoka; Sekiguchi, Kana; Ogata, Tetsuya.

Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Vol. 9887 LNCS Springer Verlag, 2016. p. 283-290 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9887 LNCS).

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

Sasaki, K, Yamakawa, M, Sekiguchi, K & Ogata, T 2016, Classification of photo and sketch images using convolutional neural networks. in Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. vol. 9887 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9887 LNCS, Springer Verlag, pp. 283-290, 25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016, Barcelona, Spain, 16/9/6. https://doi.org/10.1007/978-3-319-44781-0_34

Sasaki K, Yamakawa M, Sekiguchi K, Ogata T. Classification of photo and sketch images using convolutional neural networks. In Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Vol. 9887 LNCS. Springer Verlag. 2016. p. 283-290. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-44781-0_34

Sasaki, Kazuma ; Yamakawa, Madoka ; Sekiguchi, Kana ; Ogata, Tetsuya. / Classification of photo and sketch images using convolutional neural networks. Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Vol. 9887 LNCS Springer Verlag, 2016. pp. 283-290 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{4fa9d6bdb945491ead77a72af0a11dc6,

title = "Classification of photo and sketch images using convolutional neural networks",

abstract = "Content-Based Image Retrieval (CBIR) system enables us to access images using only images as queries, instead of keywords. Photorealistic images, and hand-drawn sketch image can be used as a queries as well. Recently, convolutional neural networks (CNNs) are used to discriminate images including sketches. However, the tasks are limited to classifying only one type of images, either photo or sketch images, due to the lack of a large dataset of sketch images and the large difference of their visual characteristics. In this paper, we introduce a simple way to prepare training datasets, which can enable the CNN model to classify both types of images by color transforming photo and illustration images. Through the training experiment, we show that the proposed method contributes to the improvement of classification accuracy.",

keywords = "Content based image retrieval, Hand-drawn sketch",

author = "Kazuma Sasaki and Madoka Yamakawa and Kana Sekiguchi and Tetsuya Ogata",

year = "2016",

doi = "10.1007/978-3-319-44781-0_34",

language = "English",

isbn = "9783319447803",

volume = "9887 LNCS",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "283--290",

booktitle = "Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings",

address = "Germany",

note = "25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016 ; Conference date: 06-09-2016 Through 09-09-2016",

}

TY - GEN

T1 - Classification of photo and sketch images using convolutional neural networks

AU - Sasaki, Kazuma

AU - Yamakawa, Madoka

AU - Sekiguchi, Kana

AU - Ogata, Tetsuya

PY - 2016

Y1 - 2016

N2 - Content-Based Image Retrieval (CBIR) system enables us to access images using only images as queries, instead of keywords. Photorealistic images, and hand-drawn sketch image can be used as a queries as well. Recently, convolutional neural networks (CNNs) are used to discriminate images including sketches. However, the tasks are limited to classifying only one type of images, either photo or sketch images, due to the lack of a large dataset of sketch images and the large difference of their visual characteristics. In this paper, we introduce a simple way to prepare training datasets, which can enable the CNN model to classify both types of images by color transforming photo and illustration images. Through the training experiment, we show that the proposed method contributes to the improvement of classification accuracy.

AB - Content-Based Image Retrieval (CBIR) system enables us to access images using only images as queries, instead of keywords. Photorealistic images, and hand-drawn sketch image can be used as a queries as well. Recently, convolutional neural networks (CNNs) are used to discriminate images including sketches. However, the tasks are limited to classifying only one type of images, either photo or sketch images, due to the lack of a large dataset of sketch images and the large difference of their visual characteristics. In this paper, we introduce a simple way to prepare training datasets, which can enable the CNN model to classify both types of images by color transforming photo and illustration images. Through the training experiment, we show that the proposed method contributes to the improvement of classification accuracy.

KW - Content based image retrieval

KW - Hand-drawn sketch

UR - http://www.scopus.com/inward/record.url?scp=84988373762&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84988373762&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-44781-0_34

DO - 10.1007/978-3-319-44781-0_34

M3 - Conference contribution

AN - SCOPUS:84988373762

SN - 9783319447803

VL - 9887 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 283

EP - 290

BT - Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings

PB - Springer Verlag

T2 - 25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016

Y2 - 6 September 2016 through 9 September 2016

ER -

Access to Document

10.1007/978-3-319-44781-0_34