Inter-Concept Distances Based on Web-scaled Image Instances

2014 Fiscal Year Final Research Report

• Project/Area Number: 24650039
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Media informatics/Database
• Research Institution: Osaka University
• Principal Investigator: BABAGUCHI Noboru 大阪大学, 工学(系)研究科(研究院), 教授 (30156541)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 概念間距離 / 画像例 / ウェブ画像 / Word Net / 画像特徴

Outline of Final Research Achievements

One of the major problems in keyword-based image retrieval is the intention gap between user's search intent and input keywords. To relax the intention gap, in this research, a couple of distance measures between concepts described by textual keywords were defined and experimentally analyzed, using a large scale of tagged-images. A system for automatically collecting tagged-images in the web space and storing the images into the database, and algorithms of measuring inter-concept distances were developed. These research results contribute to the advancement of object recognition and image retrieval.

Free Research Field

メディア情報学