Rui Li

Rui Li

Postdoctoral Researcher

College of Computing and Data Science, Nanyang Technological University

Research Interests

  • Visual Reasoning
  • Embodied AI
  • 3D/4D Reconstruction

I am a Postdoctoral Researcher at MMLab@NTU, work with Prof. Xingang Pan.
My research focuses on Embodied AI, 3D/4D Reconstruction, and Visual Reasoning, with a particular interest on inferring rich physical concepts, such as occlusion, dynamics, and interaction, from limited visual input.
Previously, I was a visiting Ph.D. student at the CVL, ETH Zürich, working with Dr. Federico Tombari and Prof. Luc Van Gool, and the GenAI Center, KAUST working with Prof. Peter Wonka. I was a Research Scientist Intern at Meta Reality Labs in Zurich.

News

Publications

* denotes equal contribution, † denotes project lead.

LaRI
LaRI: Layered Ray Intersections for Single-view 3D Geometric Reasoning
Rui Li, Biao Zhang, Zhenyu Li, Federico Tombari, Peter Wonka
International Conference on Machine Learning (ICML), 2026
Project arXiv Code Demo Dataset
TL;DR: A single-view geometric reasoning method that models unseen 3D surfaces using layered point maps, unifying object- and scene-level tasks.
Topo3DCorr
Topology-aware Feature Propagation for Unsupervised Non-rigid Point Cloud Correspondence
Haozhe Chen*, Rui Li*, Zhengbao Wang, Xinhao Zhu, Linjie Li, Tianyu Xiong, Xuan Ouyang, Jiaqi Yang
Computer Vision and Pattern Recognition (CVPR), 2026
Paper
TL;DR: A non-rigid point cloud correspondence method that learns topological structures as additional correspondence cues through self-supervision.
Sparse2DGS
Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views
Jiang Wu, Rui Li, Yu Zhu, Rong Guo, Jinqiu Sun, Yanning Zhang
Computer Vision and Pattern Recognition (CVPR), 2025
arXiv Code
TL;DR: A 3D reconstruction method that produces a watertight mesh from as few as three input images within minutes.
KYN
Know Your Neighbors: Improving Single-View Reconstruction via Spatial Vision-Language Reasoning
Rui Li, Tobias Fischer, Mattia Segu, Marc Pollefeys, Luc Van Gool, Federico Tombari
Computer Vision and Pattern Recognition (CVPR), 2024
Project arXiv Code
TL;DR: A single-view 3D reconstruction method that disambiguates occluded scene geometry by utilizing Vision-Language semantics and spatial reasoning.
GoMVS
GoMVS: Geometrically Consistent Cost Aggregation for Multi-View Stereo
Jiang Wu*, Rui Li*, Haofei Xu, Wenxun Zhao, Yu Zhu, Jinqiu Sun, Yanning Zhang
Computer Vision and Pattern Recognition (CVPR), 2024
Project arXiv Code
TL;DR: A multi-view stereo approach with geometrically consistent matching cost aggregation using monocular normals. 1st place on Tanks and Temples (Advanced) leaderboard.
DyMultiDepth
Learning to Fuse Monocular and Multi-view Cues for Multi-frame Depth Estimation in Dynamic Scenes
Rui Li, Dong Gong, Wei Yin, Hao Chen, Yu Zhu, Kaixuan Wang, Xiaozhi Chen, Jinqiu Sun, Yanning Zhang
Computer Vision and Pattern Recognition (CVPR), 2023
Project Video arXiv Code
TL;DR: A multi-frame depth estimation approach that handles dynamic areas by fusing monocular and multi-view cues in a mask-free manner.
SemDepth
Learning Depth via Leveraging Semantics: Self-supervised Monocular Depth Estimation with Both Implicit and Explicit Semantic Guidance
Rui Li, Danna Xue, Shaolin Su, Xiantuo He, Qing Mao, Yu Zhu, Jinqiu Sun, Yanning Zhang
Pattern Recognition (PR), 2023
Paper
TL;DR: A semantic-guided self-supervised depth estimation method that conducts implicit/explicit semantic guidance for high-quality and sharp depth.
Enhancing
Enhancing Self-supervised Monocular Depth Estimation via Incorporating Robust Constraints
Rui Li, Xiantuo He, Yu Zhu, Xianjun Li, Jinqiu Sun, Yanning Zhang
ACM International Conference on Multimedia (ACM MM), 2020
Paper
TL;DR: A self-supervised depth estimation method that incorporates robust constraints to improve photometric supervision.

Academic Services