Tzu-Yi Hung

Delta Air Lines (United States), Delta Electronics (China)

Papers

Total Citations

129

H-Index

About

Tzu-Yi Hung is a leading researcher in 3D and 4D computer vision, with a focus on point cloud processing for autonomous driving, robotics, and intelligent transportation. His most influential work, **SpSequenceNet** (2020, 108 citations), pioneered semantic segmentation on 4D point clouds—dynamic sequences of 3D frames—enabling scene understanding for real-world applications like autonomous navigation. He further advanced instance segmentation on 3D point clouds by introducing **Learning Regional Purity** (2022, 15 citations), a method that improves object-level recognition through region-based learning. In his latest work, **Graph Optimality-Aware Stochastic LiDAR Bundle Adjustment** (2025, 6 citations), Hung addresses large-scale LiDAR bundle adjustment, refining sensor orientation and point cloud accuracy for building high-precision navigation maps used in logistics and smart mobility. His contributions bridge the gap between static 3D analysis and dynamic 4D perception, with direct impact on autonomous systems. Hung’s research is characterized by its practical relevance, combining theoretical rigor with applications in robotics and transportation, making him a notable figure in the evolution of point cloud-based scene understanding.