文章基本信息

标题：Harvesting collective Images for Bi-Concept exploration
本地全文：下载
作者：B.Nithya priya ; K .P.kaliyamurthie
期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2013
卷号：2
期号：4
页码：1374-1382
出版社：Shri Pannalal Research Institute of Technolgy
摘要：Noised positive as well as instructive pessimistic research examples commencing the communal web, to become skilled at bi-concept detectors beginning these examples, and to apply them in a search steam engine for retrieve bi-concepts in unlabeled metaphors. We study the activities of our bi-concept search engine using 1.2 M social-tagged metaphors as a data source. Our experiments designate that harvest examples for bi-concepts differs from inveterate single-concept method, yet the examples can be composed with high accurateness using a multi-modal approach. We find that straight erudition bi-concepts is superior than oracle linear fusion of single- concept detectors Searching for the co- occurrence of two visual concepts in unlabeled images is an important step towards answering composite user queries. Traditional illustration search methods use combinations of the confidence scores of individual concept detectors to tackle such queries. Here introduce the belief of bi- concepts, an innovative concept-based retrieval method that is straightforwardly learned from social-tagged metaphors. As the number of potential bi-concepts is gigantic, physically collecting training examples is infeasible. Instead, we propose a compact disk skeleton to collect de, with a relative improvement of 100%. This study reveals the potential of learning high-order semantics from collective images, for free, suggesting promising new lines of research.
关键词：bi-concept; semantic catalog; ; visual investigate.