Baiqi Lai
Papers
2
Total Citations
17
H-Index
2
About
Baiqi Lai is a researcher at the forefront of 3D computer vision, specializing in cross-domain feature learning for 2D-3D matching. Their work addresses the fundamental challenge of aligning 2D images with 3D point clouds—a critical capability for applications in robotics, augmented reality, and autonomous navigation. Lai’s most notable contribution, the 2D3D-MVPNet, introduces a novel framework that leverages multi-view projections of point clouds to learn robust, cross-domain feature descriptors. This approach, detailed in their 2022 paper (14 citations), significantly improves matching accuracy by bridging the representational gap between 2D and 3D data. Earlier foundational work (2021, 3 citations) established the use of hard triplet loss and spatial transformer networks to enhance descriptor discriminability. By enabling reliable correspondence between visual and geometric modalities, Lai’s research pushes the boundaries of scene understanding and object recognition. Their innovative combination of projection-based learning and metric learning techniques has quickly gained traction, marking Lai as an emerging leader in the field of 3D vision and multi-modal perception.
Research Focus
Key Achievements
Top Papers
- 1
- 2