De la détection d'évènements sonores violents par SVM dans les films

Cédric Penet 1, 2 Claire-Hélène Demarty 3 Guillaume Gravier 1 Patrick Gros 1
1 TEXMEX - Multimedia content-based indexing
IRISA - Institut de Recherche en Informatique et Systèmes Aléatoires, Inria Rennes – Bretagne Atlantique
Abstract : This article studies the behaviour of a state-of-the-art support vector machine audio event detection approach, applied to violent event detection in movies. The events we are trying to detect are screams, gunshots, explosions. Contrary to others studies, we show that the state-of-theart approach does not lead to good results on this task. A study on the repartition of samples into subsets in a cross validation protocol helps explain those results and highlights a generalisation problem due to a polymorphism of considered classes. This polymorphism is demonstrated by the computation the divergence between the samples of the test database and the training database.
Document type :
Conference papers
Complete list of metadatas

Cited literature [16 references]  Display  Hide  Download

https://hal.inria.fr/inria-00595480
Contributor : Peter Sturm <>
Submitted on : Tuesday, May 24, 2011 - 7:22:18 PM
Last modification on : Friday, November 16, 2018 - 1:27:47 AM
Long-term archiving on : Thursday, August 25, 2011 - 2:28:02 AM

File

paper15.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : inria-00595480, version 1

Citation

Cédric Penet, Claire-Hélène Demarty, Guillaume Gravier, Patrick Gros. De la détection d'évènements sonores violents par SVM dans les films. ORASIS - Congrès des jeunes chercheurs en vision par ordinateur, INRIA Grenoble Rhône-Alpes, Jun 2011, Praz-sur-Arly, France. ⟨inria-00595480⟩

Share

Metrics

Record views

499

Files downloads

409