Presentation 2011-06-07
A Metadata Generation Method by Supervised Machine Learning for Telop Data in TV Videos
Yousuke WATANABE, Yutaka KATSUYAMA, Satoshi NAOI, Haruo YOKOTA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The increase of the number of video data creates a demand for scene search on video data. Our research group has been developing a system to retrieve scenes related to the specified keywords. The system generates metadata of video data by extracting telop text and appearance time from the video. In this paper, we consider not only telop text and appearance time but also telop color and position. And, we propose a metadata generation method using a machine learning algorithm. Based on usages of telops, we define classes of telops such as closed-caption and title. A telop classifier is generated from training telop data we classified. To estimate usage of unclassified telops, the classifier derives their class labels.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Video search / telop / machine learning
Paper # DE2011-24,PRMU2011-55
Date of Issue

Conference Information
Committee PRMU
Conference Date 2011/5/30(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) A Metadata Generation Method by Supervised Machine Learning for Telop Data in TV Videos
Sub Title (in English)
Keyword(1) Video search
Keyword(2) telop
Keyword(3) machine learning
1st Author's Name Yousuke WATANABE
1st Author's Affiliation Global Scientific Information and Computing Center, Tokyo Institute of Technology()
2nd Author's Name Yutaka KATSUYAMA
2nd Author's Affiliation Fujitsu Laboratory Ltd.
3rd Author's Name Satoshi NAOI
3rd Author's Affiliation Fujitsu Laboratory Ltd.
4th Author's Name Haruo YOKOTA
4th Author's Affiliation Graduate School of Information Science and Engineering, Tokyo Institute of Technology
Date 2011-06-07
Paper # DE2011-24,PRMU2011-55
Volume (vol) vol.111
Number (no) 77
Page pp.pp.-
#Pages 6
Date of Issue