Yuxuan Mu (Matthew)
Currently a graduate student at University of Alberta, Vision and Learning Lab, working on 3D computer vision. I primarily focus on 3D human motion modeling advised by Prof. Li Cheng.
I am going to join GrUVi Lab at Simon Fraser University as a Ph.D. student advised by Prof. Jason Peng.
Starting from March 2023, I also worked on 3D reconstruction and generation with Dr. Juwei Lu and Dr. Xinxin Zuo at Noah’s Ark Lab, Huawei Canada, as a research intern. My works are published/accepted in ICLR'24 and ECCV'24.
Email /
GitHub /
LinkedIn /
X
|
|
Research
My works involve 3D human motion estimation, generation, physics simulation and rendering, aiming to eventually augment our reality by building digital dynamic world clone.
'*' indicates equal contribution.
|
|
GSD: View-Guided Gaussian Splatting Diffusion for 3D Reconstruction
Yuxuan Mu, Xinxin Zuo, Chuan Guo, Yilin Wang, Juwei Lu, Xiaofei Wu, Songcen Xu, Peng Dai, Youliang Yan, Li Cheng
Accepted for ECCV, 2024
webpage /
paper /
The first 3D diffusion model directly upon Gausssian Splatting for real‐world object reconstruction, with fine-grained view-guided conditioning.
|
|
MoMask: Generative Masked Modeling of 3D Human Motions
Chuan Guo*, Yuxuan Mu*, Muhammad Gohar Javed*, Sen Wang, Li Cheng
CVPR, 2024
webpage /
paper /
code /
demo /
Star
We introduce MoMask, a novel masked modeling frame work for text‐driven 3D human motion generation with a hierarchical quantization scheme.
|
|
Generative Human Motion Stylization in Latent Space
Chuan Guo*, Yuxuan Mu*, Xinxin Zuo, Peng Dai, Youliang Yan, Juwei Lu, Li Cheng
ICLR, 2024
webpage /
paper /
code /
We propose a flexible motion style extraction and injection method from a generative perspective to solve the motion stylization task with probabilistic style space.
|
|
RACon: Retrieval-Augmented Simulated Character Locomotion Control
Yuxuan Mu, Shihao Zou, Kangning Yin, Zheng Tian, Li Cheng, Weinan Zhang, Jun Wang
ICME (Oral), 2024
paper /
We introduce an end-to-end hierarchical reinforcement learning method utilizes a task-oriented learnable retriever, a motion controller and a retrieval-augmented discriminator.
|
|
Event‐based Human Pose Tracking by Spiking Spatiotemporal Transformer
Shihao Zou, Yuxuan Mu, Xinxin Zuo, Sen Wang, Li Cheng
Arxiv Preprint, 2023
paper /
code /
Our SNNs approach uses at most 19.1% of the computation and 3.6% of the energy costs consumed by the existing methods while achieves superior performance.
|
|