Robotics and Computer Vision Lab
📧 E-mail: shlee@rcv.sejong.ac.kr / 2seung0708@naver.com / 2seung0708@gmail.com
📂 Github: https://github.com/2seung0708
🔗 LinkedIn: Seunghyeon Lee
💻 Skills
Python / C / MATLAB
Pytorch, Keras
ROS2
Research Interests
My research focuses on enhancing the robustness of deep learning–based computer vision systems under diverse real-world environmental conditions. I study instance segmentation, visual localization, 6D pose estimation, 3D detection and affordance-based perception for interactive robotics, with the goal of achieving reliable recognition and contextual understanding in complex real-world scenarios.
Education
BS: School of Intelligent Mechatronics Engineering, Sejong University (2019.03 ~ 2023.02)
Academic Excellence Scholarship for 1 semester(Fall 2019)
National Science & Technology Scholarship for 4 semester (Spring 21 – Fall 2022)
Integrated M.S.–Ph.D. Program: Department of Intelligent Mechatronics Engineering, Sejong University, (2023.03~)
Robotics and Computer Vision Lab (2021.03~ )
Publications
Crop Growth Monitoring System in Vertical Farms Based on Region-of-Interest Prediction
Yujin Hwang , Seunghyeon Lee, Taejoo Kim , Kyeonghoon Baik and Yukyung Choi*
Agriculture 2022, 12(5) 656. (Apr.2022) [paper]
Pseudo-RGB based Place Recognition through Thermal to RGB Image Translation
Seunghyeon Lee, Taejoo Kim and Yukyung Choi*
The Journal of Korea Robotics Society (JKRS) 2023, 18(1) 48-52. (Feb.2023) [paper]
6-DOF Object Pose Estimation in Aerosol Conditions: Benchmark Dataset and Baseline
Heejin Yang, Seunghyeon Lee, Taejoo Kim and Yukyung Choi*
The Journal of Institute of Control, Robotics, and Systems 2024, 30(6) 614 -620. (Jun.2024) [paper]
FRESH: Fusion-Based 3D Apple Recognition via Estimating Stem Direction Heading
Geonhwa Son, Seunghyeon Lee and Yukyung Choi*
Agriculture 2024, 14(12) 2161. (Nov.2024) [paper]
RA6D: Reliability-Aware 6D Pose Estimation via Attention-Guided Point Cloud in Aerosol Environments
Woojin Son, Seunghyeon Lee, Taejoo Kim, Geonhwa Son and Yukyung Choi*
Robotics 2026, 15(1) 8. (Dec.2025) [paper]
Patents
Method and apparatus for 3D apple detection through stem direction prediction
(가지 방향 예측을 통한 3차원 사과 검출 방법 및 그 장치)
Yukyung Choi, Geonhwa Son and Seunghyeon Lee
Korea patent ( No.10-2025-0034101)
Method and Apparatus for Reconstructing Intent Based on Multi-modal Inverse Reasoning
(멀티모달 역추론 기반 의도 재구성 방법 및 장치)
Yukyung Choi, Jaechan Lee, Seunghyeon Lee and Taejoo kim
Korea patent ( No.10-2025-0210897)
Project
Project: Crop Growth Monitoring Automation System (2021.03~2022.03)
Developing algorithms based on deep learning to monitor crop growth conditions on vertical farms
Recognize the area of the crop through instance segmentation
Development of a self-supervised learning method
Project: Environmentally robust location awareness systems (2022.03~2023.02)
Development of a strong position recognition system using thermal imaging
A Visual localization study is conducted using thermal images that are robust to day and night and weather changes
Location recognition is performed by translating thermal images into pseudo-RGB to utilize a large RGB database that has been established
Project: Estimating the pose of objects applicable to disaster situations (2023.03~2025.12)
Data set construction and algorithm study for conducting aerosol-tolerant 6d Pose Estimation research
Create a dataset by simulating aerosol situations using steam
Study robust pose estimation algorithms under visibility degradation conditions
Project: 3D Perception of Apples Considering Branch Orientation (2023.06~2024.05)
Research on 3D detection for crop harvesting with consideration of apple branch orientation
Develop 3D detection methods that account for the directional structure of apple stem
Process and curate datasets for reliable 3D detection in real agricultural environments
Project: Affordance-Based Perception for Interactive Robotics (2025.06~)
Research on perception technologies enabling robots to understand and interact with their surroundings
Object region recognition considering robot–object interaction
Algorithm development leveraging task-related contextual information