Ward@TrecVid 2016 surveillance event detection

Du, X; Yang, Y; Zhou, X

Ward@TrecVid 2016 surveillance event detection

Du, X Yang, Y

Zhou, X

Permalink

Publication Type:: Conference Proceeding
Citation:: 2016 TREC Video Retrieval Evaluation, TRECVID 2016, 2016
Issue Date:: 2016-01-01

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download Published versionAdobe PDF (104.63 kB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Du, X
dc.contributor.author	Yang, Y https://orcid.org/0000-0002-0512-880X
dc.contributor.author	Zhou, X
dc.date.accessioned	2023-02-05T23:43:32Z
dc.date.available	2023-02-05T23:43:32Z
dc.date.issued	2016-01-01
dc.identifier.citation	2016 TREC Video Retrieval Evaluation, TRECVID 2016, 2016
dc.identifier.uri	http://hdl.handle.net/10453/165908
dc.description.abstract	In this year’s competition [1], we made three changes based on last year’s system [2]. The first change was using original videos instead of resized videos. The second change was training the code books under separated cameras. The final change was performing inner vector normalization after feature encoding. In the cross validation, some of the proposed changes had impact on the detection and together made the fusion more accurate. However, in the competition, this year’s submission was inferior to the last year’s submission. We investigated the submission and found that it was caused by using the original videos, which was proposed for the intention of increasing the motion saliency in the features. As a consequence, we will not use the original videos to extract features for the next round.
dc.language	en
dc.relation.ispartof	2016 TREC Video Retrieval Evaluation, TRECVID 2016
dc.rights	info:eu-repo/semantics/openAccess
dc.title	Ward@TrecVid 2016 surveillance event detection
dc.type	Conference Proceeding
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
utslib.copyright.status	open_access	*
dc.date.updated	2023-02-05T23:43:31Z
pubs.publication-status	Published

Abstract:

In this year’s competition [1], we made three changes based on last year’s system [2]. The first change was using original videos instead of resized videos. The second change was training the code books under separated cameras. The final change was performing inner vector normalization after feature encoding. In the cross validation, some of the proposed changes had impact on the detection and together made the fusion more accurate. However, in the competition, this year’s submission was inferior to the last year’s submission. We investigated the submission and found that it was caused by using the original videos, which was proposed for the intention of increasing the motion saliency in the features. As a consequence, we will not use the original videos to extract features for the next round.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/165908