Zhirui Gao

👤 Biography

Hi there👋, my name is Zhirui Gao (in Chinese: 高志锐). I am a third-year Ph.D. student of National University of Defense Technology, supervised by Prof. Kai Xu, Prof. Wei Chen and Assoc.Prof. Renjiao Yi. From 2017 to 2021, I earned my Bachelor's degree from China University of Geosciences. I am interested in 3D vision and graphics， including 3D reconstruction, neural rendering (NeRF/3DGS), and AIGC.

I'm always open to research discussions and collaborations. In addition, I am looking for job opportunities in 3D generation and visual perception. Feel free to 📲contact me if you are interested.

📢 News

• [2025-07] Our paper BoxFusion is accepted by Pacific Graphics 2025.
• [2025-06] Two papers are accepted by ICCV2025! (PartGS and CurveGaussian).
• [2025-02] Our paper PoseProbe is accepted by IEEE TSCVT.
• [2023-12] Our paper Fdc-nerf is accepted by ICASSP (Oral), thanks for Huachen.
• [2023-06] Our paper 2d3d-matr is accepted by ICCV2023!
• [2023-04] Our paper Delving Into Crispness is accepted by TIP!
• [2023-03] Our paper NEF is accepted by CVPR2023!
• [2023-01] Our paper Deep Template Matching is accepted by IEEE CVMJ (IF:18.3)!

📑 Selected Publications [Google Scholar]

PG 2025	BoxFusion: Reconstruction-Free Open-Vocabulary 3D Object Detection via Real-Time Multi-View Box Fusion Yuqing Lan, Chenyang Zhu, Zhirui Gao, Jiazhao Zhang, Yihan Cao, Renjiao Yi, Yijie Wang, Kai Xu [arXiv] [Project Page] [Code] We propose a novel reconstruction-free paradigm for online open-vocabulary 3D object detection, which models structural object layouts with desirable running and memory efficiency.
ICCV 2025	Curve-Aware Gaussian Splatting for 3D Parametric Curve Reconstruction Zhirui Gao, Renjiao Yi, Yaqiao Dai, Xuening Zhu, Wei Chen, Chenyang Zhu, Kai Xu [arXiv] [Project Page] [Code] We present an end-to-end method for reconstructing 3D parametric curves directly from multi-view edge maps. Contrasting with the existing reconstruct-and-fit pipelines, our one-stage approach optimizes 3D parametric curves directly from 2D edge maps.
ICCV 2025	Self-supervised Learning of Hybrid Part-aware 3D Representations of 2D Gaussians and Superquadrics Zhirui Gao, Renjiao Yi, Yuhang Huang, Wei Chen, Chenyang Zhu, Kai Xu [arXiv] [Project Page] [Code] We introduce PartGS, a self-supervised part-aware reconstruction framework that integrates 2D Gaussians and superquadrics to parse objects and scenes into an interpretable decomposition, leveraging multi-view image inputs to uncover 3D structural information..
IEEE TCSVT 2025	Generic Objects as Pose Probes for Few-shot View Synthesis Zhirui Gao, Renjiao Yi, Chenyang Zhu, Ke Zhuang, Wei Chen, Kai Xu [arXiv] [Project Page] [Code] We propose to utilize everyday objects, commonly found in both images and real life, as "pose probes" to tackle few-view (3~6 unposed images) NeRF reconstruction.
ICCV 2023	2D3D-MATR: 2D-3D Matching Transformer for Detection-free Registration between Images and Point Clouds Minhao Li, Zheng Qin, Zhirui Gao, Renjiao Yi, Chenyang Zhu, Yulan Guo, Kai Xu [arXiv] [Project Page] [Code] The commonly adopted detect-then-match approach to registration finds difficulties in the cross-modality cases due to the incompatible keypoint detection and inconsistent feature description. We propose, 2D3D-MATR, a detection-free method for accurate and robust registration between images and point clouds.
CVPR 2023	NEF: Neural Edge Fields for 3D Parametric Curve Reconstruction from Multi-view Images Yunfan Ye, Renjiao Yi, Zhirui Gao, Chenyang Zhu, Zhiping Cai, Kai Xu [arXiv] [Project Page] [Code] We study the problem of reconstructing 3D feature curves of an object from a set of calibrated multi-view images. To do so, we learn a neural implicit field representing the density distribution of 3D edges which we refer to as Neural Edge Field (NEF). Inspired by NeRF, NEF is optimized with a view-based rendering loss where a 2D edge map is rendered at a given view and is compared to the ground-truth edge map extracted from the image of that view.
CVMJ 2024	Learning Accurate Template Matching with Differentiable Coarse-to-Fine Correspondence Refinement Zhirui Gao, Renjiao Yi, Zheng Qin, Yunfan Ye, Chenyang Zhu, Kai Xu [arXiv] [Project Page] [Code] Template matching is a fundamental task in computer vision and has been studied for decades. We propose an accurate template matching method based on differentiable coarse-to-fine correspondence refinement. We use an edge-aware module to overcome the domain gap between the mask template and the grayscale image, allowing robust matching.
IEEE TIP 2023	Delving into Crispness: Guided Label Refinement for Crisp Edge Detection Yunfan Ye, Renjiao Yi, Zhirui Gao, Zhiping Cai, Kai Xu [arXiv] [Project Page] [Code] We find that label quality is more important than model design to achieving crisp edge detection. We propose an iterative Canny-guided refinement of human-labeled edges whose result can be used to train crisp edge detectors.

🎓 Educations

• Ph.D. student, College of Computer Science, National University of Defense Technology
  2023 - present
  
  
• Master, College of Computer Science, National University of Defense Technology
  2021 - 2022
  
  
• B.Eng., College of Computer Science, China University of Geosciences
  2017 - 2021
  
  
  • GPA: 4.08/5.00; Average Score: 91.5/100
  • Comprehensive Ranking: 1/151

🏆 Honors & Awards

• National Scholarship
  2025
• First Prize Freshman Scholarship
  2023
• Outstanding Graduate of China University of Geosciences, Top 5%
  2021
• Outstanding Degree Thesis of China University of Geosciences, Top 5%
  2021
• National Encouragement Scholarship, Top 5%
  2019,2020
• Silver Medal of ACM-ICPC
  2019
• National Scholarship, Ministry of Education (the highest honor scholarship in China)
  2018

📝 Services

• Conference Reviewer: ICCV, MM, AAAI
• Journal Reviewer: TCSVT