|
 
|
[SFB14] Fast Cascaded Action Localization In Video Using Frame Alignment
Atelier, Poster ou Démonstration dans une Conférence Internationale : International Workshop on Computational Intelligence for Multimedia Understanding (IWCIM) 2014, October 2014, pp.-,Mots clés: action localization, video processing, learning, SVM
Résumé:
Locating human actions in videos is challenging because of
the complexity and variability of human motions, as well as
of the amount of video data to be searched. Different actions
can be composed of similar short motions and only differ by
their temporal ordering or relative durations. We propose a
method that detects and locates a set of actions in a video
database by taking into account their temporal structure at the
frame level. While other methods aggregate frames into ac-
tion parts, we leverage the complementarity between aggre-
gation and frame level comparison of sequences. With the
aim to address large scale retrieval, we introduce a two-level
cascade. The first level employs inexpensive aggregation to
filter out a large part of the video. The second level of the
cascade applies to the remaining sequences more discrimi-
nant comparisons using frame alignment. Evaluations on the
“Smoking and Drinking†and “MSR Action II†datasets show
state of the art results, as well as efficient detection and low
storage requirements.
Equipe:
vertigo
Collaboration:
LaBRI
BibTeX
| @inproceedings { | |||
| SFB14, | |||
| title | = | "{Fast Cascaded Action Localization In Video Using Frame Alignment}", | |
| author | = | " A. Stoian and M. Ferecatu and J. Benois-Pineau and M. Crucianu ", | |
| booktitle | = | "{International Workshop on Computational Intelligence for Multimedia Understanding (IWCIM) 2014}", | |
| year | = | 2014, | |
| month | = | "October", | |
| pages | = | "-", | |
| } | |||
