Presentation 2015-03-20
Specific Object Search Based On Result Re-ranking Using Region-of-interest Information
Masaya MURATA, Hidehisa NAGANO, Takahito KAWANISHI, Kaoru HIRAMATSU, Kunio KASHINO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper we first explain our retrieval approach for the specific object search task based on image queries. We then propose a methodology to effectively utilize the region-of-interest (ROI) information that is supposed to also input to the search system. Our proposed method is a search result re-ranking and the top K results in the initial search result ranking ordered by using image features within the ROI are re-ranked by using image features outside the ROI. This new approach is expected to alleviate a topic drift induced by the background information in the image query and to contribute for the improvement in the search accuracy compared with the standard approach that utilizes the background information simultaneously with the specific object information. The experimental evaluation using the TRECVID instance search task dataset shows the ameliorated search precision.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Specific object search / Region-of-interest (ROI) / Search result re-ranking / Instance search / TRECVID
Paper # BioX2014-80,PRMU2014-200
Date of Issue

Conference Information
Committee PRMU
Conference Date 2015/3/12(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Specific Object Search Based On Result Re-ranking Using Region-of-interest Information
Sub Title (in English)
Keyword(1) Specific object search
Keyword(2) Region-of-interest (ROI)
Keyword(3) Search result re-ranking
Keyword(4) Instance search
Keyword(5) TRECVID
1st Author's Name Masaya MURATA
1st Author's Affiliation NTT Communication Science Laboratories()
2nd Author's Name Hidehisa NAGANO
2nd Author's Affiliation NTT Communication Science Laboratories
3rd Author's Name Takahito KAWANISHI
3rd Author's Affiliation NTT Communication Science Laboratories
4th Author's Name Kaoru HIRAMATSU
4th Author's Affiliation NTT Communication Science Laboratories
5th Author's Name Kunio KASHINO
5th Author's Affiliation NTT Communication Science Laboratories
Date 2015-03-20
Paper # BioX2014-80,PRMU2014-200
Volume (vol) vol.114
Number (no) 521
Page pp.pp.-
#Pages 5
Date of Issue