Being-M0.5: A Real-Time Controllable Vision-Language-Motion Model

[Website] [arXiv]

We present HuMo100M, the largest multimodal motion dataset to date, including 5 million motions with fine-grained, long-form, and part-level motion labels. Leveraging our million-scale dataset HuMo100M, we present Being-M0.5, the first real-time, controllable vision-language-motion model (VLMM) that achieves diverse natural language instruction following, flexible pose initialization, long-term motion sequence generation, handling of unseen motion patterns, and precise part-aware motion control. More Visualization can be found on our [Website].

Code

We will release our code and part of our dataset soon.

Citation

If you find our work useful, please consider citing us!

@inproceedings{cao2025real,
  title={A Real-Time Controllable Vision-Language-Motion Model},
  author={Cao, Bin and Zheng, Sipeng and Wang, Ye and Xia, Lujie and Wei, Qianshan and Jin, Qin and Liu, Jing and Lu, Zongqing},
  booktitle={International Conference on Computer Vision (ICCV)},
  year={2025}
}

@inproceedings{wang2025scaling,
  title={Scaling Motion Generation Models with Million-Level Human Motions},
  author={Wang, Ye and Zheng, Sipeng and Cao, Bin and Wei, Qianshan and Zeng, Weishuai and Jin, Qin and Lu, Zongqing},
  booktitle={International Conference on Machine Learning (ICML)},
  year={2025}
}