Presentation | 2023-02-21 Classification of illustration images by mood using a combination of supervised and unsupervised learning Keisuke Kubota, Masahiro Okuda, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | With the spread of animation, games, and animated movies, users are increasingly drawing and posting illustrations on social networking services (SNS), and users are viewing illustrations more frequently. The ambiguous "atmosphere" of illustrations is thought to influence user preferences, and if illustrations could be classified according to atmosphere, they would be useful for recommendation and search. However, it is difficult to assign clear class labels to ambiguous "moods," and it is difficult to apply conventional label-based classification as is. In addition, there are many cases in which low-level features such as colors and edges of an image are similar, but the atmosphere of the image is different. Therefore, in this study, pseudo-labels that indirectly contribute to the classification of the atmosphere are assigned, and feature vectors are obtained by learning with a CNN. We then propose a method for clustering images based on the feature vectors using the K-means method. Experimental results show that the proposed method can achieve more human-like classification than conventional methods on a human-classified dataset. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | machine learning / multilabel classification / image classification / clustering / dataset |
Paper # | ITS2022-45,IE2022-62 |
Date of Issue | 2023-02-14 (ITS, IE) |
Conference Information | |
Committee | IE / ITS / ITE-MMS / ITE-ME / ITE-AIT |
---|---|
Conference Date | 2023/2/21(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Hokkaido Univ. |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Image Processing, etc. |
Chair | Kazuya Kodama(NII) / Masahiro Fujii(Utsunomiya Univ.) / Kenji Machida(NHK) / Hiroyuki Arai(Nippon Inst. of Tech.) / Hisaki Nate(Tokyo Polytechnic Univ.) |
Vice Chair | Hiroyuki Bandoh(NTT) / Toshihiko Yamazaki(Univ. of Tokyo) / Kohei Ohno(Meiji Univ.) / Naohisa Hashimoto(AIST) / / Shogo Muramatsu(Niigata Univ) |
Secretary | Hiroyuki Bandoh(KDDI Research) / Toshihiko Yamazaki(Nagoya Inst. of Tech.) / Kohei Ohno(Toyama Prefectural Univ.) / Naohisa Hashimoto(NIT, Tsuruoka College) / (Fukuoka Univ.) / Shogo Muramatsu(NHK) / (Hokkaido Univ.) |
Assistant | Shunsuke Iwamura(NHK) / Shinobu Kudo(NTT) / Taishi Swabe(NAIST) / Keiji Jimi(Gunma Univ.) |
Paper Information | |
Registration To | Technical Committee on Image Engineering / Technical Committee on Intelligent Transport Systems Technology / Technical Group on Multi-media Storage / Technical Group on Media Engineering / Technical Group on Artistic Image Technology |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Classification of illustration images by mood using a combination of supervised and unsupervised learning |
Sub Title (in English) | |
Keyword(1) | machine learning |
Keyword(2) | multilabel classification |
Keyword(3) | image classification |
Keyword(4) | clustering |
Keyword(5) | dataset |
1st Author's Name | Keisuke Kubota |
1st Author's Affiliation | Doshisha University(Doshisha Univ.) |
2nd Author's Name | Masahiro Okuda |
2nd Author's Affiliation | Doshisha University(Doshisha Univ.) |
Date | 2023-02-21 |
Paper # | ITS2022-45,IE2022-62 |
Volume (vol) | vol.122 |
Number (no) | ITS-384,IE-385 |
Page | pp.pp.19-24(ITS), pp.19-24(IE), |
#Pages | 6 |
Date of Issue | 2023-02-14 (ITS, IE) |