Abstract: This paper addresses the challenging task of generating dynamic Human-Object Interactions from textual descriptions, named Text2HOI. While most existing works assume interactions with ...