2 datasets found

Generating diverse and natural 3D human motions from text.

Text-to-motion synthesis aims to generate 3D human motion that not only precisely reflects the textual description but reveals the motion details as much as possible.