Papers
5
Total Citations
431
H-Index
4
About
Wenxiu Sun is a leading researcher in 3D computer vision, with a focus on point cloud processing, 3D object reconstruction, and depth completion. Her most impactful contribution is the GRNet (Gridding Residual Network), introduced in her 2020 paper, which has garnered over 375 citations. This work addresses the critical challenge of dense point cloud completion from incomplete 3D data—a key problem in vision and robotics. Unlike mainstream methods that use MLPs and often lose structural details, GRNet innovatively grids point clouds into 3D voxel grids, enabling convolutional neural networks to capture fine-grained geometric details, achieving state-of-the-art results. Sun has also advanced 3D object reconstruction from stereo images, proposing methods that overcome the limitations of template-matching approaches to improve generalization and reconstruction quality. Her recent work includes leading the MIPI 2023 Challenge on RGB+ToF depth completion, demonstrating her leadership in combining RGB and sparse Time-of-Flight measurements for robust depth estimation. With her pioneering work in point cloud completion and 3D reconstruction, Sun has established herself as a key figure in pushing the boundaries of 3D vision for real-world applications.
Research Focus
Key Achievements
Top Papers
- 1GRNet: Gridding Residual Network for Dense Point Cloud Completion375 citations · 2020
- 2GRNet: Gridding Residual Network for Dense Point Cloud Completion29 citations · 2020
- 3Toward 3D object reconstruction from stereo images19 citations · 2021
- 4MIPI 2023 Challenge on RGB+ToF Depth Completion: Methods and Results6 citations · 2023
- 5Toward 3D Object Reconstruction from Stereo Images2 citations · 2019