Presentation 2008-02-21
High-level feature extraction from video using tree clustering
Taichi NAKAMURA, Koichi SHINODA, Sadaoki FURUI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a novel method using a tree-structured codebook and a node selection technique for high-level feature extraction from video data. We first quantize local SIFT descriptors to construct a tree-structred codebook shared among all the high-level features. Then, we select nodes to be used as visual words for each high-level feature. We also propose motion words which are defined as visual words with motion activities. We evaluate the proposed method using TRECVID 2006 and 2007 database. By selecting nodes, Mean Average Precision(MAP) improved from 0.097 (baseline) to 0.100. By using the motion word, MAP father improved to 0.102. These results confirmed the effectiveness of the proposed method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Video annotation / High-level feature extraction / Object recognition / SIFT descriptor / Tree-structured clustering
Paper # PRMU2007-220
Date of Issue

Conference Information
Committee PRMU
Conference Date 2008/2/14(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) High-level feature extraction from video using tree clustering
Sub Title (in English)
Keyword(1) Video annotation
Keyword(2) High-level feature extraction
Keyword(3) Object recognition
Keyword(4) SIFT descriptor
Keyword(5) Tree-structured clustering
1st Author's Name Taichi NAKAMURA
1st Author's Affiliation Department of Computer Sience, Tokyo Institute of technology()
2nd Author's Name Koichi SHINODA
2nd Author's Affiliation Department of Computer Sience, Tokyo Institute of technology
3rd Author's Name Sadaoki FURUI
3rd Author's Affiliation Department of Computer Sience, Tokyo Institute of technology
Date 2008-02-21
Paper # PRMU2007-220
Volume (vol) vol.107
Number (no) 491
Page pp.pp.-
#Pages 6
Date of Issue