Haozhe Xie (谢浩哲)
Research Fellow at MMLab@NTU
College of Computing and Data Science
Nanyang Technological University
Emails:
- General Inquiries: root [at] haozhexie [dot] com
- Academic Matters: academic [at] haozhexie [dot] com
Biography
I am currently working as a research fellow at MMLab@NTU, Nanyang Technological University, working with Prof. Ziwei Liu.
Previously, I was a Senior Research Scientist at Tencent AI Lab from 2021 to 2023.
I received my Ph.D. from the VILab at Harbin Institute of Technology in 2021, supervised by Prof. Hongxun Yao. During my Ph.D., I also interned at SenseTime Research, mentored by Dr. Wenxiu Sun.
My research focuses on computer vision, with a particular emphasis on 3D vision and robotics.
Selected Publications
MonoArt: Progressive Structural Reasoning for Monocular Articulated 3D Reconstruction
arXiv 2603.19231
HSImul3R: Physics-in-the-Loop Reconstruction of Simulation-Ready Human–Scene Interactions
arXiv 2603.15612
Bridging Semantic and Kinematic Conditions with Diffusion-based Discrete Motion Tokenizer
arXiv 2603.19227
InfiniteDance: Scalable 3D Dance Generation Towards in-the-wild Generalization
arXiv 2603.13375
DynamicVLA: A Vision-Language-Action Model for Dynamic Object Manipulation
arXiv 2601.22153
- Project Page
- Code
- Stars
Compositional Generative Model of Unbounded 4D Cities
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) 48.1 (2026): 312-328
- Project Page
- Code
- Stars
CrowdMoGen: Zero-Shot Text-Driven Collective Motion Generation
International Journal of Computer Vision (IJCV) 134.1 (2026): 29
Generative Gaussian Splatting for Unbounded 3D City Generation
CVPR 2025
- Project Page
- Code
- Demo
- Stars
3DTopia-XL: Scaling High-quality 3D Asset Generation via Primitive Diffusion
CVPR 2025
- Project Page
- Code
- Demo
- Stars
DynamicCity: Large-Scale 4D Occupancy Generation from Dynamic Scenes
ICLR 2025
- Project Page
- Code
- Stars
CityDreamer: Compositional Generative Model of Unbounded 3D Cities
CVPR 2024
- Project Page
- Code
- Demo
- Stars
3D Scene and Object Reconstruction from Multiple Sources and Viewpoints
PhD Thesis, Harbin Institute of Technology, 2021
Pix2Vox++: Multi-scale Context-aware 3D Object Reconstruction from Single and Multiple Images
International Journal of Computer Vision (IJCV) 128.12 (2020): 2919-2935
- Project Page
- Code
- Stars
Pix2Vox: Context-aware 3D Reconstruction from Single and Multi-view Images
ICCV 2019
- Project Page
- Code
- Stars
Research Experience
Research Fellow
Mar 2023 - Present | MMLab@NTU, Nanyang Technological University
Working with Prof. Ziwei Liu.
Senior Research Scientist
Aug 2021 - Mar 2023 | Tencent AI Lab
Working with Dr. Hong Shang.
Outstanding Contibutor (2022H1) & Excellent Individual (2022H2)
Research Intern
Mar 2019 - Nov 2020 | SenseTime Research
Mentored by Dr. Wenxiu Sun.
Outstanding Intern (2019H2)
Invited Talks
- Toward World Models: From 3D to 4D City Generation. 3D Computer Vision (3D 视觉工坊), Online. March 2025. [Video playback]
- 3D Object and Scene Reconstruction Meets Neural Networks. Faculty of Computing, Harbin Institute of Technology, Harbin, China. May 2022. [News]
- Deep Learning Fundamentals. Zhejiang Institute of Research and Innovation, The University of Hong Kong, Hangzhou, China. July 2019. [News]
- Computer Fundamentals. Hangzhou No.14 High School, Hangzhou, China. August 2014.
- Data Structure. Hangzhou No.14 High School, Hangzhou, China. August 2012.
Academic Services
- Co-organizer: CVPR'25 Tutorial: From Video Generation to World Model
- Aera Chair: ICLR
- Conference Reviewer: CVPR, ICCV, ECCV, ICML, ICLR, NeurIPS, SIGGRAPH Asia, AAAI, Eurographics, 3DV, WACV, ACM Multimedia
- Journal Reviewer: TPAMI, IJCV, TVCG, TOG, TIP, TMM, PR, CSUR
Teaching
- NTU AI6126: Advanced Computer Vision, Teaching Assistant, Spring 2025.
- HIT CS32261: Audio-Visual Signal Processing, Teaching Assistant, Fall 2018.