Dong-Guw Lee
Papers
2
Total Citations
4
H-Index
2
About
Dong-Guw Lee is a rising researcher at the intersection of computer vision and robotics, whose work addresses fundamental perception challenges in autonomous systems. His primary research areas include visual place recognition, 3D scene understanding, and robotic manipulation, with a particular focus on handling difficult visual domains. Lee’s notable contributions include pioneering night-to-day thermal image translation for deep thermal place recognition, enabling robust localization under adverse lighting conditions where conventional methods fail. His most impactful work, "TranSplat: Surface Embedding-Guided 3D Gaussian Splatting for Transparent Object Manipulation" (2025), tackles the long-standing problem of robotic perception of transparent objects—a challenge that has stymied the field due to depth sensor failures. By introducing surface embedding-guided 3D Gaussian splatting, Lee’s approach achieves accurate dense depth completion and enables reliable manipulation of transparent objects, a breakthrough with significant implications for industrial automation and service robotics. Though early in his career, with his key papers already garnering citations, Lee’s work demonstrates a clear trajectory toward solving some of the most persistent perceptual bottlenecks in robotics, marking him as a promising innovator in the field.
Research Focus
Key Achievements
Top Papers
- 1
- 2