Mining visual actions from movies - Archive ouverte HAL Access content directly
Conference Papers Year : 2009

Mining visual actions from movies

(1, 2) , (3) , (1)
1
2
3

Abstract

This paper presents an approach for mining visual actions from real-world videos. Given a large number of movies, we want to automatically extract short video sequences corresponding to visual human actions. Firstly, we retrieve actions by mining verbs extracted from the transcripts aligned with the videos. Not all of these samples visually characterize the action and, therefore, we rank these videos by visual consistency. We investigate two unsupervised outlier detection methods: one-class Support Vector Machine (SVM) and densest component estimation of a similarity graph. Alternatively, we show how to use automatic weak supervision provided by a random background class, either by directly applying a binary SVM, or by using an iterative re-training scheme for Support Vector Regression machines (SVR). Experimental results explore actions in 144 episodes of the TV series ''Buffy the Vampire Slayer'' and show: (a) the applicability of our approach to a large scale set of real-world videos, (b) the importance of visual consistency for ranking videos retrieved from text, (c) the added value of random non-action samples and (d) the ability of our iterative SVR re-training algorithm to handle weak supervision. The quality of the rankings obtained is assessed on manually annotated data for six different action classes.
Fichier principal
Vignette du fichier
gaidon_mining_actions_bmvc2009.pdf (1.6 Mo) Télécharger le fichier
Vignette du fichier
top_punch_02.png (452.01 Ko) Télécharger le fichier
Vignette du fichier
one_pager_mining_actions_bmvc09.pdf (1.25 Mo) Télécharger le fichier
Vignette du fichier
poster_buffy_BMVC09.png (11.87 Mo) Télécharger le fichier
Vignette du fichier
top_10_fall.avi (7.74 Mo) Télécharger le fichier
Vignette du fichier
top_10_get_up.avi (7.12 Mo) Télécharger le fichier
Vignette du fichier
top_10_walk.avi (4.7 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Format : Figure, Image
Format : Other
Format : Other
Format : Other
Format : Other
Format : Other
Loading...

Dates and versions

inria-00440973 , version 1 (14-12-2009)
inria-00440973 , version 2 (25-04-2012)

Identifiers

Cite

Adrien Gaidon, Marcin Marszalek, Cordelia Schmid. Mining visual actions from movies. British Machine Vision Conference, British Machine Vision Association, Sep 2009, Londres, United Kingdom. pp.125.1-125.11, ⟨10.5244/C.23.125⟩. ⟨inria-00440973v2⟩
640 View
828 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More