3D Human Motion Generation
Abstract
Human motion generation aims to produce natural, realistic and diverse human movements suitable for a wide range of applications, covering film production, video games, AR/VR, human-robot interaction, and digital human representation. In this talk, I will present the current methodologies employed in human motion generation. In addition, I will present our recent research efforts focused on the challenging task of human reaction generation, which is focused on generating appropriate reactions in response to given actions. Finally, I will conclude by presenting new results demonstrating the use of diffusion models for the generation of human interactions.