Conference item icon

Conference item

Thread-safe: towards recognizing human actions across shot boundaries

Abstract:

We study the task of recognizing human actions in video whilst paying attention to the shot and thread editing structure. Most existing action recognition algorithms ignore this structure, but it is generally present in edited TV and film material.

To this end, we make the following contributions: first, we introduce a new dataset of human actions to study the occurrence/reoccurrence of patterns of human actions in edited TV material; second, we propose composing a video into threads of related shots, removing some of the discontinuities due to shot boundaries; and third, we show the benefits of utilizing video threads in recognizing human actions. The experiments demonstrate that human action retrieval accuracy can be improved using threads.

Publication status:
Published
Peer review status:
Peer reviewed

Actions


Access Document


Publisher copy:
10.1007/978-3-319-16817-3_15

Authors


More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Engineering Science
Oxford college:
Brasenose College
Role:
Author
ORCID:
0000-0002-8945-8573


More from this funder
Funder identifier:
https://ror.org/0439y7842
Grant:
EP/I012001/1
More from this funder
Funder identifier:
https://ror.org/03wnrjx87


Publisher:
Springer
Host title:
Computer Vision -ACCV 2014
Pages:
222-237
Series:
Lecture Notes in Computer Science
Series number:
9006
Publication date:
2015-04-17
Acceptance date:
2014-11-01
Event title:
12th Asian Conference on Computer Vision (ACCV 2014)
Event location:
Singapore
Event start date:
2014-11-01
Event end date:
2014-11-05
DOI:
EISSN:
1611-3349
ISBN:
9783319168166


Language:
English
Keywords:
Pubs id:
572816
Local pid:
pubs:572816
Deposit date:
2024-07-12

Terms of use



Views and Downloads






If you are the owner of this record, you can report an update to it here: Report update to this record

TO TOP