| 大会名称 |
|---|
| 2010年 情報科学技術フォーラム(FIT) |
| 大会コ－ド |
| F |
| 開催年 |
| 2010 |
| 発行日 |
| 2010/8/20 |
| セッション番号 |
| 2H |
| セッション名 |
| 画像重要度 |
| 講演日 |
| 2010/09/07 |
| 講演場所(会議室等) |
| H会場（総合学習プラザ2F 第12講義室） |
| 講演番号 |
| I-009 |
| タイトル |
| Attention-based Clustering for Grouping Photo Collections by Quality of Composition |
| 著者名 |
| デシルヴァ ガムヘワゲ, 相澤 清晴, |
| キーワード |
| Photo Quality, Saliency _Map, Image Clustering, Multimedia Retrieval |
| 抄録 |
| We propose a system for automated clustering search results in a large photo collections according the quality of photographic composition. The system uses computational visual attention as a representative for composition. We analyze features derived from saliency maps of a set of training images to identify natural groupings of images. Supervised learning based on these clusters enables grouping large collections of images according to composition. The results of preliminary experiments, conducted on reviewed photographs from a stock photography archive and images uploaded to Flickr, show that the system is capable of meaningfully grouping the photos. |
| 本文pdf |
| PDF download (559.1KB) |