CV
Contact Information
| Name | Xiaohu Zhang |
| Professional Title | Ph.D. |
| xiaohuzhang@zju.edu.cn, shelf@tencent.com | |
| Location | Shenzhen, Guangdong 518000 |
Professional Summary
My research focuses on integrating deep learning with traditional geometry processing algorithms to build efficient, interpretable Neural-Analytical Hybrid Pipelines for robust discrete modeling compatible with downstream tasks such as animation and physical simulation.
Experience
-
2026 - present Shenzhen, China
Researcher on 3D AIGC (Qingyun Program)
Tencent IEG
Research on 3D AIGC for games.
-
2024 - 2025 Hangzhou, China
Mesh Deep learning Intern
Shining 3D
Developed an intrinsic deep feature point extraction module for dental applications.
Education
-
2020 - 2026 Hangzhou, China
Ph.D.
Zhejiang University
Computer Science
- Focus: Computer Graphics, Deep Geometry Processing
-
2016 - 2020 Hangzhou, China
Bachelor
Zhejiang University
Polymer Science
Publications
-
2026 Learning Sparse Singularities for Cross Field Design
ACM Trans. Graph.
To enable problem-aware quad meshing, we explore learning the singularity structure to encode quad topology for enhanced generalization. Compared to analytical approaches, our method automatically integrates problem priors. Furthermore, compared to cross-field learning techniques, our method adapts better to various topologies.
-
2026 Versatile Curve Design by Level Set With Quadratic Convergence
IEEE Transactions on Visualization and Computer Graphics.
We propose a smooth approximation framework for level set based curve design on manifolds. Our framework enables the use of Newton’s method and allows for highly flexible controls.