Jiaxin Wei

I am a second-year PhD student in the Smart Robotics Lab at the Technical University of Munich, supervised by Prof. Stefan Leutenegger. My research focuses on 3D vision, mapping, and object-level perception. I am also passionate about bridging the gap between the physical and virtual worlds. I earned my Master's degree from ShanghaiTech University, where I worked in the Mobile Perception Lab under the supervision of Prof. Laurent Kneip.

Email  /  Scholar  /  LinkedIn  /  Github

Research

A real-time mapping system that simultaneously generates a TSDF grid and a photo-realistic 3D map for various downstream tasks.

A novel extension of the classical PnP algorithm that estimates object pose without requiring exact 3D coordinates by handling unknown anisotropic scaling.

A decoupled pipeline for RGB-only object pose estimation that separately estimates metric size and 6D pose, achieving superior rotation accuracy.

A novel sampling model inspired by spherical multi-view scanners that enables compact and efficient point cloud representation for 3D shape completion.

A fine-grained CAD model retrieval in large-scale databases by combining learned shape descriptors with geometry-based re-ranking.

Source code from Jon Barron's website.