search

Aubrey-ao / HumanBehaviorAnimation

178 stars 15 forks source link

Model inference speed #13

Open cyk990422 opened 8 months ago

cyk990422 commented 8 months ago

Hello, Dr. Tenglong I would like to ask about the generation speed of GestureDiffuCLIP and RhythmicGesticulator, such as how many seconds it takes to generate a slice, and how many seconds this slice is a gesture. Do they have the potential to be applied to real-time tasks? I'm very much looking forward to your answer. I'm very interested in your article.

Aubrey-ao commented 7 months ago

Hi，

Thx for your attention. Both systems can be optimized to a real-time performance (The inference strategy of the diffusion model can refer to the latest Stable Diffusion model).

PS: The core of training a good-quality gesture generation model is Data. A curated dataset is more important than the architecture : )