Hung-Ting Su
Papers
3
Total Citations
37
H-Index
2
About
Hung-Ting Su is an emerging researcher at the intersection of computer vision, depth estimation, and embodied AI, with work that advances how machines perceive and interact with complex three-dimensional environments. His most notable contribution, "S³: Learnable Sparse Signal Superdensity for Guided Depth Estimation" (2021, 21 citations), introduces an innovative approach to dense depth estimation by addressing a fundamental challenge in the field: the low density and imbalance inherent in sparse sensing modalities such as LiDAR and Radar. By developing a learnable framework to enhance sparse signal representation, Su's method meaningfully improves depth estimation quality across applications in robotics, 3D reconstruction, and augmented reality. His collaborative work on OCID-Ref (2021, 14 citations), presented at NAACL, further demonstrates his breadth of expertise, contributing a novel 3D robotic dataset with embodied language grounding designed to evaluate visual grounding in cluttered, real-world environments like offices and warehouses. This dataset directly addresses the underexplored challenge of identifying occluded objects through natural language, pushing the boundaries of human-robot interaction research. Across his work, Su demonstrates a consistent commitment to bridging perception and practical robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3