Dong-Guw (David) Lee
Ph.D. in Mechanical Engineering, Seoul National University
Thermal infrared perception, robot vision, and vision-language models
I am a robotics and computer vision researcher working on thermal infrared perception for robot autonomy. I completed my Ph.D. in Mechanical Engineering at Seoul National University, advised by Prof. Ayoung Kim in the Robust Perception and Mobile Robotics Laboratory.
My research focuses on data scaling and radiometric representation learning for deep thermal perception, including thermal object detection, RGB-to-thermal translation, thermal image denoising, place recognition, SLAM-oriented perception, and vision-language prompting for controllable image translation. I also work on underwater robotic monitoring, open-vocabulary perception, image enhancement, and biofouling segmentation for infrastructure inspection.
Thermal perception for robots operating beyond visible-light assumptions.
Thermal Infrared Vision
Radiometric representation learning, tone mapping, denoising, and deep perception pipelines for thermal cameras.
Robot Perception
Object detection, segmentation, place recognition, and SLAM-adjacent perception for field robots and infrastructure monitoring.
Generative Translation
RGB-to-thermal translation, controllable visual-language prompting, and synthetic data generation for rare operating conditions.
- Project lead for underwater robotic monitoring of floating infrastructure, including open-vocabulary segmentation, dehazing, mosaicing, and at-sea robot trials.
- Project lead for thermal image-based navigation and shipyard monitoring with synthetic thermal data generation and real-time rare-class detection.
- Organizer for the Workshop on Thermal Infrared in Robotics at ICRA 2025.
Prior to SNU, I completed an M.S. in the Robotics Program at KAIST, supervised by Prof. Dong-Soo Kwon, and a BEng in Electrical and Electronic Engineering at the University of Bristol.
selected publications
- Thermal Chameleon: Task-Adaptive Tone-Mapping for Radiometric Thermal-Infrared ImagesIEEE Robotics and Automation Letters, 2024
- Night-to-day thermal image translation for deep thermal place recognitionIntelligent Service Robotics, 2023
- Thera: Thermal-Aware Visual-Language Prompting for Controllable RGB-to-Thermal Infrared TranslationIn IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
- Edge-Guided Multi-Domain RGB-to-TIR Image Translation for Training Vision Tasks with Challenging LabelsIn Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), May 2023