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 proceedingConference 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 languageEnglish
    Title of host publicationArtificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings
    PublisherSpringer Verlag
    Pages283-290
    Number of pages8
    Volume9887 LNCS
    ISBN (Print)9783319447803
    DOIs
    Publication statusPublished - 2016
    Event25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016 - Barcelona, Spain
    Duration: 2016 Sep 62016 Sep 9

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume9887 LNCS
    ISSN (Print)03029743
    ISSN (Electronic)16113349

    Other

    Other25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016
    CountrySpain
    CityBarcelona
    Period16/9/616/9/9

    Fingerprint

    Neural Networks
    Neural networks
    Image retrieval
    Color
    Experiments
    Query
    Content-based Image Retrieval
    Large Data Sets
    Neural Network Model
    Classify
    Experiment

    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 proceedingConference 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",

    }

    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

    ER -