MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation

Nhat M. Hoang^1,2*  Gong Kehong²⁺  Chuan Guo^2*  Michael Bi Mi² 

¹Nanyang Technological University  ²Huawei Technologies Co., Ltd 

^*Work done while interning at Huawei ⁺Corresponding author

This repository contains the official implementation of MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation.
Please refer to our project page for more visualizations.

[02/2024] Release training and inference code for the model MDM (text-to-motion, action-to-motion), MotionDiffuse (text-to-motion), and EDGE (music-to-dance).

[01/2024] Paper uploaded to arXiv.

[12/2023] Paper accepted at AAAI-24 Main Track! 🎉

This is a research implementation and, in general, will not be regularly updated or maintained long after release.

If you find our work useful for your research and development, please consider citing the paper:

@misc{hoang2024motionmix,
  title={MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation}, 
  author={Nhat M. Hoang and Kehong Gong and Chuan Guo and Michael Bi Mi},
  year={2024},
  eprint={2401.11115},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}