Yuliang Xiu
Assistant Professor at Westlake University
Hangzhou City, Zhejiang, China
Summary
🤩
Rockstar🎓
Top SchoolYuliang Xiu is an assistant professor and virtual avatar researcher specializing in 3D human digitalization, bridging computer vision and graphics with a focus on reconstruction, modeling, animation, and 2D pose tracking. With 11 years' experience and a PhD from Tübingen under Michael J. Black, he leads the Endless AI Lab at Westlake University after internships at Meta's Codec Avatar Lab and industry R&D roles. He has made notable open-source contributions to CVPR-recognized projects like ICON and ECON, implementing voxel layers, mesh upsampling, and semantic-aware improvements that power animatable, clothed human reconstructions. Comfortable across full-stack research code and ML engineering, he combines rigorous academic training with practical systems work—from ResNet-based pose notebooks to production-ready mesh processing. Colleagues describe him as someone who plays with pixels and voxels to create believable digital humans while keeping an eye on deployment and animation fidelity.
11 years of coding experience
3 years of employment as a software developer
Exchange student, Computer Science and Information Engineering, Exchange student, Computer Science and Information Engineering at National Cheng Kung UniversityDoctor of Philosophy - PhD, Computer Vision and Graphics, Doctor of Philosophy - PhD, Computer Vision and Graphics at University of TübingenBachelor of Engineering - BE, Digital Media Technology, Bachelor of Engineering - BE, Digital Media Technology at Shandong UniversityMaster of Science - MS, Computer Science, Master of Science - MS, Computer Science at Shanghai Jiao Tong University
Chinese, English
Github Skills (12)
computer-vision10
pytorch10
deep-learning10
python10
pose-estimation10
3d-reconstruction10
image-processing9
data-augmentation9
machine-learning9
heatmap8
computer-graphics8
open3d7
Programming languages (8)
Github contributions (5)
YuliangXiu/MobilePose
Jan 2018 - Mar 2022
Light-weight Single Person Pose Estimator
Role in this project:
ML Engineer
ML EngineerContributions:86 commits, 13 PRs, 66 pushes in 4 years 3 months
Contributions summary:Yuliang primarily worked on a Jupyter Notebook demonstrating the use of ResNet-18 for pose estimation. They initialized the notebook, imported necessary libraries, and set up the environment including GPU configuration. The user defined functions to rescale and transform image data, and implemented a custom dataset class for handling pose data. They also defined a Net class using ResNet-18, configured the loss function, and set up the optimizer for training. The commits indicate an effort to visualize pose estimation results, using the visualization code.
shufflenetv2realtimemobilenetsqueezenetshufflenet
YuliangXiu/ECON
Dec 2022 - Mar 2023
[CVPR'23, Highlight] ECON: Explicit Clothed humans Optimized via Normal integration
Role in this project:ML Engineer
ML EngineerContributions:1 review, 59 commits, 18 PRs in 3 months
Contributions summary:Yuliang contributed significantly to the ECON repository, focusing on 3D human reconstruction and avatar generation. Their work involved modifying the `apps/infer.py` file to include semantic-aware hand and face replacement, integrating mesh upsampling and subdivision operations, and incorporating part removal techniques based on SMPL-X. Further contributions included the implementation of NICP (Nearest Iterative Closest Point) for SMPL-X completion, enhancing the animatable avatar capabilities within the project. The user also integrated color querying from RGB images and Open3D, demonstrating an understanding of mesh processing and computer vision techniques.
arxivmetaversesmpl-bodycomputer-graphicscomputer-vision
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.