Aesthetic Quality Classification of Photographs

• Item type: SIG Technical Reports(1)
• 公開日: 2011-05-12
• タイトル: Aesthetic Quality Classification of Photographs
• タイトル / 言語: en
• タイトル: Aesthetic Quality Classification of Photographs
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: D論セッション
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: The University of Tokyo／Toshiba Corporation
• 著者所属: The University of Tokyo
• 著者所属(英): en
• 著者所属(英): The University of Tokyo / Toshiba Corporation
• 著者所属(英): en
• 著者所属(英): The University of Tokyo
• 著者名: Masashi, Nishiyama; Yoichi, Sato
• 著者名(英): Masashi, Nishiyama; Yoichi, Sato
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: The aesthetic quality of a photograph is a significant cue for inferring the level of appreciation that arises when people view digital content on various types of computer systems. It would be useful to classify the aesthetic quality in photographs to more easily edit large photograph collections that are available because of the widespread use of digital cameras and the Internet. The research community is currently tackling this challenging problem of aesthetic quality classification, which automatically assesses whether a photograph generates aesthetic appreciation. This paper focuses on developing techniques for aesthetic quality classification of photographs. In particular, we discuss detecting of multiple-subject regions and assessing color harmony in order to enhance the performance of aesthetic quality classification. These techniques play an important role in determining aesthetic quality. We statistically build these quality classifiers using large photograph databases collected on websites where users manually provide quality labels to photographs. Furthermore, we demonstrate that our aesthetic quality classifier emulates how people edit large photograph collections.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: The aesthetic quality of a photograph is a significant cue for inferring the level of appreciation that arises when people view digital content on various types of computer systems. It would be useful to classify the aesthetic quality in photographs to more easily edit large photograph collections that are available because of the widespread use of digital cameras and the Internet. The research community is currently tackling this challenging problem of aesthetic quality classification, which automatically assesses whether a photograph generates aesthetic appreciation. This paper focuses on developing techniques for aesthetic quality classification of photographs. In particular, we discuss detecting of multiple-subject regions and assessing color harmony in order to enhance the performance of aesthetic quality classification. These techniques play an important role in determining aesthetic quality. We statistically build these quality classifiers using large photograph databases collected on websites where users manually provide quality labels to photographs. Furthermore, we demonstrate that our aesthetic quality classifier emulates how people edit large photograph collections.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA11131797
• 書誌情報: 研究報告コンピュータビジョンとイメージメディア（CVIM）; 巻 2011-CVIM-177, 号 33, p. 1-16, 発行日 2011-05-12
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会