[论文] Bridging Semantic and Kinematic Conditions with Diffusion-based Discre...

论文概要

研究领域: CV 作者: Chenyang Gu, Mingyuan Zhang, Haozhe Xie 发布时间: 2026-03-19 arXiv: 2503.16903

中文摘要

我们提出MoTok，一种基于扩散的离散运动token化器，通过将运动恢复委托给扩散解码器来实现语义抽象与细粒度重建的解耦。在HumanML3D上，我们的方法在仅使用六分之一token的情况下显著提升了可控性和保真度，轨迹误差从0.72厘米降至0.08厘米。

原文摘要

We propose MoTok, a diffusion-based discrete motion tokenizer that decouples semantic abstraction from fine-grained reconstruction by delegating motion recovery to a diffusion decoder. On HumanML3D, our method significantly improves controllability and fidelity while using only one-sixth of the tokens, reducing trajectory error from 0.72 cm to 0.08 cm.

#论文 #arXiv #CV #小凯