ORAND at TRECVID 2015: Instance search and video hyperlinking tasks

Juan Manuel Barrios; Jose M. Saavedra; Felipe Ramirez; David Contreras

ORAND at TRECVID 2015: Instance search and video hyperlinking tasks

Juan Manuel Barrios, Jose M. Saavedra, Felipe Ramirez, David Contreras

Research output: Contribution to conference › Paper › peer-review

1 Scopus citations

Abstract

ORAND S.A. is a Chilean company focused on developing applied research in Computer Science. This report describes the participation of the ORAND team at Instance Search (INS) and Video Hyperlinking (LNK) tasks in TRECVID 2015. The INS participation consisted in four submissions to automatic detection. All the submissions used the four samples for each topic without using video queries (type A). We tested different score propagation algorithms where the based on low-level features achieved best performance. The LNK participation consisted in four submissions. The submissions considering low-level features (color histogram, edges histogram and acoustic energies) outperformed the semantic descriptors (concepts and captions).

Original language	English
State	Published - 1 Jan 2015
Externally published	Yes
Event	2015 TREC Video Retrieval Evaluation, TRECVID 2015 - Gaithersburg, United States Duration: 16 Nov 2015 → 18 Nov 2015

Conference

Conference	2015 TREC Video Retrieval Evaluation, TRECVID 2015
Country/Territory	United States
City	Gaithersburg
Period	16/11/15 → 18/11/15

Bibliographical note

Publisher Copyright:
© 2015 TRECVID.

Cite this

@conference{31351b00453f4e46ba32e6a6aebe45fb,

title = "ORAND at TRECVID 2015: Instance search and video hyperlinking tasks",

abstract = "ORAND S.A. is a Chilean company focused on developing applied research in Computer Science. This report describes the participation of the ORAND team at Instance Search (INS) and Video Hyperlinking (LNK) tasks in TRECVID 2015. The INS participation consisted in four submissions to automatic detection. All the submissions used the four samples for each topic without using video queries (type A). We tested different score propagation algorithms where the based on low-level features achieved best performance. The LNK participation consisted in four submissions. The submissions considering low-level features (color histogram, edges histogram and acoustic energies) outperformed the semantic descriptors (concepts and captions).",

author = "Barrios, {Juan Manuel} and Saavedra, {Jose M.} and Felipe Ramirez and David Contreras",

note = "Publisher Copyright: {\textcopyright} 2015 TRECVID.; 2015 TREC Video Retrieval Evaluation, TRECVID 2015 ; Conference date: 16-11-2015 Through 18-11-2015",

year = "2015",

month = jan,

day = "1",

language = "English",

}

TY - CONF

T1 - ORAND at TRECVID 2015

T2 - 2015 TREC Video Retrieval Evaluation, TRECVID 2015

AU - Barrios, Juan Manuel

AU - Saavedra, Jose M.

AU - Ramirez, Felipe

AU - Contreras, David

PY - 2015/1/1

Y1 - 2015/1/1

N2 - ORAND S.A. is a Chilean company focused on developing applied research in Computer Science. This report describes the participation of the ORAND team at Instance Search (INS) and Video Hyperlinking (LNK) tasks in TRECVID 2015. The INS participation consisted in four submissions to automatic detection. All the submissions used the four samples for each topic without using video queries (type A). We tested different score propagation algorithms where the based on low-level features achieved best performance. The LNK participation consisted in four submissions. The submissions considering low-level features (color histogram, edges histogram and acoustic energies) outperformed the semantic descriptors (concepts and captions).

AB - ORAND S.A. is a Chilean company focused on developing applied research in Computer Science. This report describes the participation of the ORAND team at Instance Search (INS) and Video Hyperlinking (LNK) tasks in TRECVID 2015. The INS participation consisted in four submissions to automatic detection. All the submissions used the four samples for each topic without using video queries (type A). We tested different score propagation algorithms where the based on low-level features achieved best performance. The LNK participation consisted in four submissions. The submissions considering low-level features (color histogram, edges histogram and acoustic energies) outperformed the semantic descriptors (concepts and captions).

UR - http://www.scopus.com/inward/record.url?scp=85204490942&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85204490942

Y2 - 16 November 2015 through 18 November 2015

ER -

ORAND at TRECVID 2015: Instance search and video hyperlinking tasks

Abstract

Conference

Bibliographical note

Other files and links

Fingerprint

Cite this