Wenxu Zhou

I am a first-year Ph.D. student at University of Science and Technology of China (USTC). I am currently conducting research in the Graphics & Geometric Computing Laboratory (GCL), School of Mathematical Sciences, under the supervision of Prof. Juyong Zhang. Before joining GCL, I received my M.S. degree in Electronic Information from the Department of Electronic Engineering and Information Science (EEIS), USTC, where I was advised by Prof. Dong Yin. I earned my B.Eng. degree in Electronic Information Engineering from the School of Electronics and Information Engineering, Anhui University (AHU).

My research interests include 3d vision, generative models and embodied AI.

Email  /  Scholar  /  CV  /  GitHub  /  Twitter  

profile photo

News

  • [Apr. 2026] Served as a reviewer for T-PAMI.
  • [Nov. 2025] One paper accepted to BIBM 2025.
  • [Sep. 2025] One paper accepted to PRCV 2025.

Research Interests

Specifically, my work (some papers are highlighted) focuses on the following sub-fields:

  • 3D/4D Reconstruction and Generation
  • World Model
  • Human-Object Interaction

Publications

IL3D: A Large-scale indoor layout dataset for LLM-Driven 3D Scene Generation
Wenxu Zhou, Kaixuan Nie, Hang Du, Dong Yin, Wei Huang, Siqiang Guo, Xiaobo Zhang, Pengbo Hu
arXiv, 2025
project page / arXiv / BibTeX / code / dataset

IL3D, a large-scale dataset with 27,816 indoor layouts and 29,215 3D assets, supports LLM-driven 3D scene generation. It includes natural language annotations and rigorous benchmarks, enhancing the 3D scene generation task.

Open-Vocabulary Endoscopic Scene Understanding via 4D Language Gaussian Splatting
Wenxu Zhou, Dong Yin
BIBM, 2025
paper / BibTeX / code

EndoLGS, a 4D Language Gaussian Splatting for endoscopic scenes dynamic reconstruction and semantic understanding, shows significant potential for robot-assisted surgery.

Endo2DGS: Endoscopic Scene Reconstruction with High-fidelity Geometry
Wenxu Zhou, Taoran Sun, Tianle Hu, Jiulin Li, Dong Yin
PRCV, 2025
paper / BibTeX

Endo2DGS, a novel reconstruction framework for endoscopic scenes, addresses geometric inconsistencies in dynamic digestive tract modeling, achieving high-fidelity 4D reconstruction.

Teaching

  • [Fall 2024] Teaching Assistant, Data Structure and Algorithm, Department of EEIS, USTC.
  • Academic Services

    Conference Reviewer: PRCV, AAAI.
    Journal Reviewer: T-PAMI.


    Web Template @ Here.