Xianming Liu

Papers

Total Citations

H-Index

About

Xianming Liu is a leading researcher in 3D computer vision and geometric deep learning, with a focus on point cloud processing and 6D object tracking. His most impactful contribution is in point cloud upsampling (PCU), where he pioneered self-supervised, arbitrary-scale implicit methods that generate dense, uniform 3D point clouds from sparse LiDAR inputs—a critical capability for autonomous driving, robotics, and AR/VR. His 2023 paper on this topic has already garnered 17 citations, reflecting its influence in addressing the practical challenge of sensor sparsity without requiring ground-truth dense data. Liu also pushes boundaries in 6D object pose tracking, as demonstrated in his 2025 work "Zero6DOT," which achieves zero-shot tracking from monocular RGB video, eliminating the need for CAD models or multi-modal sensors. This innovation promises to democratize 6D tracking for robotic manipulation and virtual reality applications. By tackling both data scarcity (via self-supervision) and resource constraints (via zero-shot learning), Liu’s research consistently advances real-world deployment of 3D vision systems. His work bridges fundamental geometry processing with practical engineering, making him a key figure in next-generation spatial AI.