文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Merging Storyboard Strategies and Automatic Retrieval for Improving Interactive Video Searc
侵权投诉

Merging Storyboard Strategies and Automatic Retrieval for Improving Interactive Video Searc

The Carnegie Mellon University Informedia group has enjoyed consistent success with TRECVID interactive search using traditional storyboard interfaces for shot-based retrieval. For TRECVID 2006 the output of automatic search was included for the first time

Merging Storyboard Strategies and Automatic Retrieval for

Improving Interactive Video Search

I

Michael G. Christel School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 USA

1-412-268-7799

christel@cs.cmu.edu

ABSTRACT

The Carnegie Mellon University Informedia group has enjoyed consistent success with TRECVD interactive search using traditional storyboard interfaces for shot-based retrieval. For TRECVID 2006 the output of automatic search was included for

the first time with storyboards, both as an option for an interactive

user and in a different run as the sole means of access. The automatic search makes use of relevance-based probabilistic retrieval models to determine weights for combining retrieval sources when addressing a given topic. Storyboard-based access using automatic search output outperformed extreme video retrieval interfaces of manual browsing with resizable pages and rapid serial visualization of keyframes that used the same output. Further, the full nformedia interface with automatic search I

results as an option along with other query mechanisms scored significantly better than all other TRECVID 2006 interactive search systems. Attributes of the automatic search and interactive search systems are discussed to further optimize shot-based retrieval from news corpora.

Categories and Subject Descriptors

H.5.1 [Information Interfaces and Presentation]: Multimedia nformation Systems – evaluation,video; H.3.3 [Information Storage and Retrieval]: Information Search and Retrieval

General Terms

Experimentation, Human Factors

Keywords

Video retrieval, TRECVID results analysis

1.INTRODUCTION

This paper reports on the Carnegie Mellon University Informedia group’s TRECVD 2006 interactive search experiences and successes, particularly the use of traditional storyboard interfaces for shot-based retrieval. In 2006 the ranked shot lists from automatic search were also provided to users, leading to a

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

CIVR'07, July 9-11, 2007, Amsterdam, The Netherlands.

Merging Storyboard Strategies and Automatic Retrieval for Improving Interactive Video Searc

Copyright 2007 ACM 978-1-59593-733-9/07/0007 ...$5.00.

Rong Yan

IBM TJ Watson Research Center Hawthorne, NY 10532 USA

1-914-784-7142

yanr@us.ibm.com

I

significant boost in performance. An introduction to TRECVID is given, followed by the motivation for the experiments reported here. Specifically, the new work here is as follows, conducted through a series of TRECVID 2006 search runs: 1.Can the results of fully automatic search help the user in interactive search?

2.

Does the method of presentation matter, i.e., is there a difference between extreme “push many images before the user’s eyes” under automatic control, versus image layout and paging in traditional storyboards under user control? 3.

Does the freedom to solicit different shot sets matter, i.e., rather than being locked in to the output of a fully automatic search with interactive filtering, let the user also generate other shot sets for inspection through querying by text, imagery, and concepts?

TRECVID at NIST has the goal to promote progress in content-based retrieval from digital video via open, metrics-based evaluation. The corpora have ranged from documentaries to advertising films to broadcast news, with international participation growing from 12 to 54 companies and academic institutions from 2001 to 2006 [6]. A number of tasks are defined in TRECVD, including shot detection, story segmentation, semantic feature extraction, and information retrieval, with this paper focused only on the retrieval task.

TRECVID video searches deal with shots, where a shot is defined as a single continuous camera operation without an editor’s cut, fade or dissolve – typically 2-10 seconds long for broadcast news. The TRECVD search task is defined as follows: given a multimedia statement of information need (topic) and the common shot reference, return a ranked list of up to 1000 shots from the reference which best satisfy the need. Success is measured based on quantities of relevant shots retrieved, in particular the metrics of recall and precision. The two are combined into a single measure of performance, average precision, which measures precision after each relevant shot is retrieved for a given topic. Average precision is then itself averaged over all of the topics to produce a mean average precision (MAP) metric for evaluating a system’s performance. Search types include “automatic” in which the query topic is taken as is with no human modifications, and “interactive” in which the user can view the topic, interact with the system, see results, and refine queries and browsing strategies interactively while pursuing a solution. The interactive user has no prior knowledge of the search test collection or topics.

第1页

你可能喜欢

热门文档

相关文档

  • ...Digital Video Retrieval System

    Merging Storyboard Strategies and Automatic Retrieval for Improving Interactive Video Search. In Proc. Image and Video Retrieval (CIVR) (July 2007). [2] ...

  • ...Digital Video Retrieval System

    Merging Storyboard Strategies and Automatic Retrieval for Improving Interactive Video Search. In Proc. Image and Video Retrieval (CIVR) (July 2007). [2] ...

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email