Emotions are an important aspect of daily life. They are expressed through different channels, some of which are visual : facial expressions, postures and body movements. In addition to speech, this is an important means of communication that can provide or capture information about a person’s emotional state. In computer science, applications seeking to take into account the emotional state of the user or seeking to produce expressive virtual characters are increasingly numerous : education, medicine, commerce, security, robotics and entertainment (cinema, video games, etc.). In their human-machine interactions, these applications try to move towards gestural and conversational interactions where expressions take an important place. In image synthesis, animated virtual characters are present everywhere in video media such as movies, video games or virtual worlds. The way these characters move, act in their world and interact is an important focus for their creators. And to make them lively and attractive, they need realistic gestures and emotions. To meet this expectation, systems must be able to analyze and understand the behaviors, i.e. the emotional state, of a person, but they must also be able to produce or assist in the production of high quality, expressive and diverse animation. Our work takes place in this perspective of understanding, recognizing and synthesizing gestures, movements, animations of the face and body producing an expression or a style. The notion of expression and style is the common thread of our work in computer vision and animation synthesis. We offer advances in facial and body expression analysis and recognition, as well as in animation synthesis and editing with a particular style or expression.