Xiuhong Lin
Papers
1
Total Citations
3
H-Index
1
About
Xiuhong Lin’s research lies at the intersection of computer vision and 3D scene understanding, with a particular focus on bridging the gap between 2D images and 3D spatial data. Her most notable contribution is the development of a novel framework for learning cross-domain descriptors that enable robust 2D-3D matching. In her 2021 paper, she introduced a hard triplet loss combined with a spatial transformer network, a method that significantly improves the accuracy of aligning 2D image features with 3D point clouds—a critical task for applications in augmented reality, robotics, and autonomous navigation. While her citation count is still growing, this work has been recognized for its innovative approach to handling the geometric and appearance variations inherent in cross-domain matching. Lin’s research demonstrates a deep understanding of both metric learning and spatial transformations, offering a practical solution to a long-standing challenge in 3D vision. Her work is particularly valuable for students and researchers seeking to understand how deep learning can be tailored to align disparate data modalities, and it lays a strong foundation for future advances in 3D object recognition and scene reconstruction.
Research Focus
Key Achievements
Top Papers
- 1