A study on object detection method from manga images using CNN

    研究成果: Conference contribution

    2 引用 (Scopus)

    抄録

    Japanese comics (manga) are popular content worldwide. In order to acquire metadata from manga images, techniques automatic recognition of manga content have been studied. Recently, Convolutional Neural Network (CNN) has been applied to object detection in manga images. R-CNN and Fast R-CNN generate region proposals by Selective Search. Faster R-CNN generates them using CNN layers called Region Proposal Network (RPN). Single Shot MultiBox Detector (SSD), the latest detection method, performs object classification and box adjustment for small regions in an image. These methods are effective to natural images. However, it is unclear whether such methods work properly to manga images or not, since those image features are different from natural images. In this paper, we examine the effectiveness of manga object detection by comparing Fast R-CNN, Faster R-CNN, and SSD. Here, manga objects are panel layout, speech balloon, character face, and text. Experimental results show that Fast R-CNN is effective for panel layout and speech balloon, whereas Faster R-CNN is effective for character face and text.

    元の言語English
    ホスト出版物のタイトル2018 International Workshop on Advanced Image Technology, IWAIT 2018
    出版者Institute of Electrical and Electronics Engineers Inc.
    ページ1-4
    ページ数4
    ISBN(電子版)9781538626153
    DOI
    出版物ステータスPublished - 2018 5 30
    イベント2018 International Workshop on Advanced Image Technology, IWAIT 2018 - Chiang Mai, Thailand
    継続期間: 2018 1 72018 1 9

    Other

    Other2018 International Workshop on Advanced Image Technology, IWAIT 2018
    Thailand
    Chiang Mai
    期間18/1/718/1/9

    Fingerprint

    Neural networks
    Balloons
    Detectors
    Object detection
    Network layers
    Metadata

    ASJC Scopus subject areas

    • Computer Networks and Communications
    • Computer Vision and Pattern Recognition
    • Media Technology

    これを引用

    Yanagisawa, H., Yamashita, T., & Watanabe, H. (2018). A study on object detection method from manga images using CNN. : 2018 International Workshop on Advanced Image Technology, IWAIT 2018 (pp. 1-4). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IWAIT.2018.8369633

    A study on object detection method from manga images using CNN. / Yanagisawa, Hideaki; Yamashita, Takuro; Watanabe, Hiroshi.

    2018 International Workshop on Advanced Image Technology, IWAIT 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 1-4.

    研究成果: Conference contribution

    Yanagisawa, H, Yamashita, T & Watanabe, H 2018, A study on object detection method from manga images using CNN. : 2018 International Workshop on Advanced Image Technology, IWAIT 2018. Institute of Electrical and Electronics Engineers Inc., pp. 1-4, 2018 International Workshop on Advanced Image Technology, IWAIT 2018, Chiang Mai, Thailand, 18/1/7. https://doi.org/10.1109/IWAIT.2018.8369633
    Yanagisawa H, Yamashita T, Watanabe H. A study on object detection method from manga images using CNN. : 2018 International Workshop on Advanced Image Technology, IWAIT 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 1-4 https://doi.org/10.1109/IWAIT.2018.8369633
    Yanagisawa, Hideaki ; Yamashita, Takuro ; Watanabe, Hiroshi. / A study on object detection method from manga images using CNN. 2018 International Workshop on Advanced Image Technology, IWAIT 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 1-4
    @inproceedings{c25ce8597fa94071950d3aea58f427d4,
    title = "A study on object detection method from manga images using CNN",
    abstract = "Japanese comics (manga) are popular content worldwide. In order to acquire metadata from manga images, techniques automatic recognition of manga content have been studied. Recently, Convolutional Neural Network (CNN) has been applied to object detection in manga images. R-CNN and Fast R-CNN generate region proposals by Selective Search. Faster R-CNN generates them using CNN layers called Region Proposal Network (RPN). Single Shot MultiBox Detector (SSD), the latest detection method, performs object classification and box adjustment for small regions in an image. These methods are effective to natural images. However, it is unclear whether such methods work properly to manga images or not, since those image features are different from natural images. In this paper, we examine the effectiveness of manga object detection by comparing Fast R-CNN, Faster R-CNN, and SSD. Here, manga objects are panel layout, speech balloon, character face, and text. Experimental results show that Fast R-CNN is effective for panel layout and speech balloon, whereas Faster R-CNN is effective for character face and text.",
    keywords = "CNN, Fast R-CNN, Faster R-CNN, Manga, Object Detection, SSD",
    author = "Hideaki Yanagisawa and Takuro Yamashita and Hiroshi Watanabe",
    year = "2018",
    month = "5",
    day = "30",
    doi = "10.1109/IWAIT.2018.8369633",
    language = "English",
    pages = "1--4",
    booktitle = "2018 International Workshop on Advanced Image Technology, IWAIT 2018",
    publisher = "Institute of Electrical and Electronics Engineers Inc.",

    }

    TY - GEN

    T1 - A study on object detection method from manga images using CNN

    AU - Yanagisawa, Hideaki

    AU - Yamashita, Takuro

    AU - Watanabe, Hiroshi

    PY - 2018/5/30

    Y1 - 2018/5/30

    N2 - Japanese comics (manga) are popular content worldwide. In order to acquire metadata from manga images, techniques automatic recognition of manga content have been studied. Recently, Convolutional Neural Network (CNN) has been applied to object detection in manga images. R-CNN and Fast R-CNN generate region proposals by Selective Search. Faster R-CNN generates them using CNN layers called Region Proposal Network (RPN). Single Shot MultiBox Detector (SSD), the latest detection method, performs object classification and box adjustment for small regions in an image. These methods are effective to natural images. However, it is unclear whether such methods work properly to manga images or not, since those image features are different from natural images. In this paper, we examine the effectiveness of manga object detection by comparing Fast R-CNN, Faster R-CNN, and SSD. Here, manga objects are panel layout, speech balloon, character face, and text. Experimental results show that Fast R-CNN is effective for panel layout and speech balloon, whereas Faster R-CNN is effective for character face and text.

    AB - Japanese comics (manga) are popular content worldwide. In order to acquire metadata from manga images, techniques automatic recognition of manga content have been studied. Recently, Convolutional Neural Network (CNN) has been applied to object detection in manga images. R-CNN and Fast R-CNN generate region proposals by Selective Search. Faster R-CNN generates them using CNN layers called Region Proposal Network (RPN). Single Shot MultiBox Detector (SSD), the latest detection method, performs object classification and box adjustment for small regions in an image. These methods are effective to natural images. However, it is unclear whether such methods work properly to manga images or not, since those image features are different from natural images. In this paper, we examine the effectiveness of manga object detection by comparing Fast R-CNN, Faster R-CNN, and SSD. Here, manga objects are panel layout, speech balloon, character face, and text. Experimental results show that Fast R-CNN is effective for panel layout and speech balloon, whereas Faster R-CNN is effective for character face and text.

    KW - CNN

    KW - Fast R-CNN

    KW - Faster R-CNN

    KW - Manga

    KW - Object Detection

    KW - SSD

    UR - http://www.scopus.com/inward/record.url?scp=85048781686&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85048781686&partnerID=8YFLogxK

    U2 - 10.1109/IWAIT.2018.8369633

    DO - 10.1109/IWAIT.2018.8369633

    M3 - Conference contribution

    AN - SCOPUS:85048781686

    SP - 1

    EP - 4

    BT - 2018 International Workshop on Advanced Image Technology, IWAIT 2018

    PB - Institute of Electrical and Electronics Engineers Inc.

    ER -