Fusical: Multimodal Fusion for Video Sentiment

Level fusion analysis of recurrent audio and video neural network for violence detection in railway

“…At the same time, research is moving towards the processing of multi-modal signals [20] and fusion approaches have been experimented in use cases like speech enhancement [21], emotion recognition [22], tracking of multiple speakers [23], action recognition [24] or scene classification [25]. Because no open synchronized audio-video dataset for transport applications exists, these techniques have not been experimented for violence detection in this environment.…”