Human Motion " Following " system using Hidden Markov Models and application to dance performance

Rémy Muller (1)

1 Analyse et synthèse sonores [Paris]

This work presents a movement “following” system based on “Hidden Markov Models” and motion descriptors extracted from video. The primary application is in performing arts such as dance but the methodology is general enough to be applied in other fields as long as appropriate descriptors are available. Different motion features, segmentations, decoding methods and data analysis such as Principal Components analysis have been performed and compared in order to show how they affect the “following” performances. Both contemporary dance performances and synthetic animations have been used in order to evaluate the system.

Mots clés: HMM

Domaines: Traitement du signal et de l'image [eess.SP]; Machine Learning [stat.ML]; Statistiques [stat]; Vision par ordinateur et reconnaissance de formes [cs.CV]

Rémy Muller. Human Motion " Following " system using Hidden Markov Models and application to dance performance. [Research Report] STMS - Sciences et Technologies de la Musique et du Son UMR 9912 IRCAM-CNRS-UPMC. 2004. ⟨hal-01556701⟩