Hello! I am a second-year Ph.D. student advised by Prof. Kuk-Jin Yoon at the Visual Intelligence Lab (VILab), Korea Advanced Institute of Science and Technology (KAIST).

My research focuses on developing methods that reliably extract and utilize semantic information from visual data. I am broadly interested in how models can maintain semantic consistency across domains, modalities, and interaction scenarios.

Recently, I have become increasingly interested in multimodal setting and representation learning, including how strong foundation models—such as Diffusion models, and SAM/SAM2-style segmentation systems—can be adapted and leveraged for robust visual understanding. I also have interest in egocentric perception and the integration of cues like eye-gaze to enhance semantic interpretation and user interaction in immersive VR/AR environments. Currently, I am exploring Multi-Agent Reinforcement Learning, which I believe is the direction I should pursue as an AI researcher.

Ultimately, my research aims to understand how semantic structures can be extracted from diverse visual modalities and effectively applied to create more adaptive and intelligent perception systems. I am always open to discussions and collaborations—feel free to reach out anytime!

My CV can be found in here

Publications

Multimodal Distribution Matching for Vision-Language Dataset Distillation
Jongoh Jeong*, Hoyong Kwon*, Minseok Kim*, Kuk-Jin Yoon
The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026

Project Page

Bootstrapping Video Semantic Segmentation Model via Distillation-assisted Test-Time Adaptation
Jihun Kim*, Hoyong Kwon*, Hyeokjun Kweon*, Kuk-Jin Yoon
The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026

ArXiv

DC-TTA: Divide-and-Conquer Framework for Test-Time Adaptation of Interactive Segmentation
Jihun Kim*, Hoyong Kwon*, Hyeokjun Kweon*, Wooseong Jeong, Kuk-Jin Yoon
International Conference on Computer Vision (ICCV) 2025

Paper | Supp | ArXiv | Code

Phase Concentration and Shortcut Suppression for Weakly Supervised Semantic Segmentation
Hoyong Kwon, Jaeseok Jeong, Sung-Hoon Yoon, Kuk-Jin Yoon
European Conference on Computer Vision (ECCV) 2024

Paper | Supp | Code

Diffusion-Guided Weakly Supervised Semantic Segmentation
Sung-Hoon Yoon, Hoyong Kwon*, Jaeseok Jeong*, Daehee Park, Kuk-Jin Yoon
European Conference on Computer Vision (ECCV) 2024

Paper | Supp | Code

Class Tokens Infusion for Weakly Supervised Semantic Segmentation
Sung-Hoon Yoon, Hoyong Kwon, Hyeonseong Kim, Kuk-Jin Yoon
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024

Paper | Supp | Code

Selected Honors and Awards

• 2023 / Magna Cum Laude, KAIST BS
• 2021-2022 / National Science & Technology Scholarship, Korea Student Aid Foundation
• 2020 / Dean’s List, KAIST College of Engineering
  

Research Experiences

• 2025.03 - Present / Project Leader / Development of Mobility and Connectivity Platform for Unmanned Autonomous Delivery / Multi-Modal Fusion, 3D Object&Lane Detection, ROS2
  
• 2024.01 - 2025.02 / Project Member / Development of Mobility and Connectivity Platform for Unmanned Autonomous Delivery / Multi-Modal Fusion, 3D Object&Lane Detection, ROS2
  
• 2022.03 - 2022.08 / Internship / NCSOFT Vision AI Lab Recognition Team / Keypoint Detection, Markerless Face Wrapping