Video processing to analyse human gesture is a hot and promising topic since it was adopted in many various applications as video surveillance, medical monitoring, virtual reality, interaction with machines, etc. The human class  is a core concern of research and industry because it points the way towards the artificial intelligence. How can a computer do the same interpretation of a scene that a human can do?  The possibilities are endless but are reliant on the effectiveness of primal processes:  analysing and recognizing human gestures and behaviours. This workshop aims to investigate the various meanings, media (2D or 3D captors) and constraints (real-time)  that leads to answer to the three mean steps of this problematic : tracking, description and learning. The processes have to be be adapted to the particularly complex class of the human. The way to follow movement, describe gesture and recognize behaviours  at different level (full body, legs, hand, etc.) will be brought.