Presentation 2011-12-16
Toward High-Performance Video Semantic Indexing
Nakamasa INOUE, Koichi SHINODA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) TokyoTech, in collaboration with Canon Inc., achieved the best performance in the Semantic Indexing task of TRECVID2011 hosted by the National Institute of Standards and Technology (NIST), outperforming 28 teams from all over the world. In this paper, we show the outline of TRECVID and the Semantic Indexing task, review our approaches in the past years, and explain our method developed this year. Our method first comprises GMM supervectors from audio-visual signals and inputs them into support vector machines to detect the features. This method successfully achieves robustness against differences in locations, movements, colors, shapes, angles, and brightness, etc. We utilize TokyoTech's supercomputer, TSUBAME, in large computation.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) video search / semantic indexing / GMM supervector / SVM
Paper # PRMU2011-140
Date of Issue

Conference Information
Committee PRMU
Conference Date 2011/12/8(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) Toward High-Performance Video Semantic Indexing
Sub Title (in English)
Keyword(1) video search
Keyword(2) semantic indexing
Keyword(3) GMM supervector
Keyword(4) SVM
1st Author's Name Nakamasa INOUE
1st Author's Affiliation Department of Computer Science, Tokyo Institute of Technology()
2nd Author's Name Koichi SHINODA
2nd Author's Affiliation Department of Computer Science, Tokyo Institute of Technology
Date 2011-12-16
Paper # PRMU2011-140
Volume (vol) vol.111
Number (no) 353
Page pp.pp.-
#Pages 6
Date of Issue