Xiangtai Li

Nanyang Technological University, Peking University

Papers

Total Citations

306

H-Index

About

Xiangtai Li is a leading researcher in computer vision, specializing in visual segmentation—the task of partitioning images, video frames, and point clouds into meaningful segments. His work bridges foundational deep learning methods with cutting-edge transformer architectures, driving advances in autonomous driving, robot sensing, and medical imaging. Li’s highly cited survey, “Transformer-Based Visual Segmentation: A Survey” (2024, 192 citations), has become a key reference for the community, synthesizing the rapid evolution of transformer-based segmentation models. He also made a notable contribution to niche yet critical applications with “Enhanced Boundary Learning for Glass-like Object Segmentation” (2021, 103 citations), which tackles the challenging problem of segmenting transparent objects like windows and bottles—essential for robot navigation and grasping. By improving boundary learning in these complex scenes, Li’s work directly addresses real-world sensing limitations. With over 300 combined citations for these papers alone, his research demonstrates both broad impact and technical depth, establishing him as a rising authority in visual perception and segmentation.