Xiaoyang Bai

301 Haking Wong Building

The University of Hong Kong

I am currently a Postdoctoral Fellow at WeLight Lab, Department of Electrical and Computer Engineering, HKU. I am fortunate to be supervised by Prof. Evan Y. Peng.

I received my Ph.D. degree in Computer Science from University of Illinois Urbana-Champaign in 2023, advised by Prof. Alexander Schwing and mentored by Prof. Viktor Gruev and Prof. David Forsyth. Before that, I completed my undergraduate study at University of California, Berkeley, where I majored in Computer Science, Applied Mathematics, Cognitive Science and Linguistics.

My research interests span the fields of Computer Vision, Computer Graphics and Computational Optics. I am particularly attracted to the idea of physical world optimization from unconventional visual modalities (event camera, polarization, hyperspectral image, etc.). My non-research interests include science (fantasy/speculative) fictions, zoos and gaming.

news

Jun 03, 2026	One paper has been accepted by TVCG.
Apr 09, 2026	Our paper PhyGaP: Physically-Grounded Gaussians with Polarization Cues have been selected as Award Candidate and Oral Presentation for CVPR 2026!
Feb 23, 2026	One paper has been accepted by CVMJ.
Feb 21, 2026	Two papers have been accepted by CVPR 2026.
Nov 15, 2025	One paper has been accepted by 3DV 2026.

selected publications

PhyGaP: Physically-Grounded Gaussians with Polarization Cues

Jiale Wu, Xiaoyang Bai, Zongqi He , and 2 more authors

In Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2026

HTML
Structure-grounded Training Strategies Aid Generalization in Stereo Matching

Liangxun Ou, Yuhui Liu, Zhenyang Li , and 2 more authors

In Proceedings of International Conference on 3D Vision , 2026

HTML
EventTracer: Fast Path Tracing-based Event Stream Rendering

Zhenyang Li, Xiaoyang Bai, Jinfan Lu , and 3 more authors

arXiv preprint arXiv:2508.18071, 2025

HTML

Enhanced Velocity Field Modeling for Gaussian Video Reconstruction

Zhenyang Li^*, Xiaoyang Bai^*, Tongchen Zhang , and 3 more authors

In Proceedings of IEEE International Symposium on Mixed and Augmented Realitys (ISMAR) , 2025

Bib HTML

@inproceedings{li2025enhanced,
  title = {Enhanced Velocity Field Modeling for Gaussian Video Reconstruction},
  author = {Li, Zhenyang and Bai, Xiaoyang and Zhang, Tongchen and Shen, Pengfei and Xu, Weiwei and Peng, Yifan},
  booktitle = {Proceedings of IEEE International Symposium on Mixed and Augmented Realitys (ISMAR)},
  year = {2025},
  organization = {IEEE},
  google_scholar_id = {qjMakFHDy7sC}
}

Learning RGBD imaging via asymmetrically focused stereo cameras

Liangxun Ou, Yuhui Liu, Xiaoyang Bai , and 1 more author

The Visual Computer, 2025

Bib HTML

@article{ou2025learning,
  title = {Learning RGBD imaging via asymmetrically focused stereo cameras},
  author = {Ou, Liangxun and Liu, Yuhui and Bai, Xiaoyang and Peng, Yifan},
  journal = {The Visual Computer},
  year = {2025},
  publisher = {Springer},
  google_scholar_id = {IjCSPb-OGe4C}
}

Learning a global underwater geolocalization model with sectoral transformer

Xiaoyang Bai, Zhongmin Zhu, Alexander Schwing , and 2 more authors

Optics Express, 2024

Bib HTML

@article{bai2024learning,
  title = {Learning a global underwater geolocalization model with sectoral transformer},
  author = {Bai, Xiaoyang and Zhu, Zhongmin and Schwing, Alexander and Forsyth, David and Gruev, Viktor},
  journal = {Optics Express},
  volume = {32},
  number = {12},
  pages = {20706--20718},
  year = {2024},
  publisher = {Optica Publishing Group},
  google_scholar_id = {2osOgNQ5qMEC}
}

Polarization-based underwater geolocalization with deep learning

Xiaoyang Bai, Zuodong Liang, Zhongmin Zhu , and 3 more authors

eLight, 2023

Bib HTML

@article{bai2023polarization,
  title = {Polarization-based underwater geolocalization with deep learning},
  author = {Bai, Xiaoyang and Liang, Zuodong and Zhu, Zhongmin and Schwing, Alexander and Forsyth, David and Gruev, Viktor},
  journal = {eLight},
  volume = {3},
  number = {1},
  pages = {15},
  year = {2023},
  publisher = {Springer},
  google_scholar_id = {u-x6o8ySG0sC}
}