Mots audio-visuels joints pour la détection de scènes violentes dans les vidéos

This paper presents an audio-visual data representation for violent scenes detection in Hollywood movies. Existing works in this field consider either the audio or the visual information; or their shallow fusion. None has yet explored their joint dependence for violent scenes detection. We propose a feature which provides strong multimodal audio and visual cues by first joining the audio and the visual features and then revealing statistically the joint multimodal patterns. Experimental validation was conducted in the context of the “Violent Scenes Detection” task of the MediaEval 2013 Multimedia benchmark. The obtained results show the potential of the proposed approach in comparison to methods using audio and visual feature separately and other fusion methods. MOTS-CLÉS : Indéxation sémantique, Analyse de contenu, Fusion audio-visuelle, Multimédia, MediaEval.