I am a Research Scientist at Electronics and Telecommunications Research Institute (ETRI).
I received Ph.D. and M.S. degrees in Electrical Engineering from the Korea Advanced Institute of Science and Technology (KAIST), where I was advised by In So Kweon. During my Ph.D. studies, I interned at Adobe Research and was honored with the Qualcomm Innovation Fellowship.
My research aims to build robust multi-modal AI models capable of understanding and generating complex real-world scenarios, with a specific focus on co-designing effective data and learning frameworks. My primary research interests include the following areas, but also open to exploring other challenging and impactful problems.
pkyong7 [at] etri.re.kr
pkyong7 [at] kaist.ac.kr
218, Gajeong-ro, Yuseong-gu, Daejeon, Republic of Korea, 34129
PhD, Major in EE, KAIST, 2023
on "Towards Universal Visual Scene Understanding in the Wild"
Advisor: Prof. In So Kweon
MS, Major in EE, KAIST, 2019
on "Learning unpaired video-to-video translation for domain adaptation"
Advisor: Prof. In So Kweon
BS, Double Major in ME and EE, KAIST, 2018
A Multimodal Chain of Tools for Described Object Detection
NeurIPS 2024 Workshop on Compositional Learning
[ Paper ]
KOALA: Empirical Lessons Toward Memory-Efficient and Fast Diffusion Models for Text-to-Image Synthesis
NeurIPS 2024
*Also presented at CVPR 2024 Workshop on "Generative Models for Computer Vision"
Media coverage: covered by YTN, Yonhap News, AI Times, and many local media
[ Paper / Project page / Code ]
Weak-to-Strong Compositional Learning from Generative Models for Language-based Object Detection
ECCV 2024
*Also presented at CVPR 2024 Workshop on "Generative Models for Computer Vision"
3rd place in the OmniLabel Challenge @ ECCV2024
[ Paper ]
MTMMC: A Large-Scale Real-World Multi-Modal Camera Tracking Benchmark
CVPR 2024
[ Paper / Project page ]
Test-time Adaptation in the Dynamic World with Compound Domain Knowledge Management
RAL-ICRA 2024
[ Paper ]
Joint Self-supervised Learning and Adversarial Adaptation for Monocular Depth Depth Estimation from Thermal Image
MVA 2023
[ Paper ]
Learning Classifiers of Prototypes and Reciprocal Points for Universal Domain Adaptation
WACV 2023
[ Paper ]
Self-supervised Monocular Depth Estimation from Thermal Images via Adversarial Multi-spectral Adaptation
WACV 2023
Received Best Student Paper Award in WACV 2023
[ Paper ]
Bridging Images and Videos: A Simple Learning Framework for Large Vocabulary Video Object Detection
ECCV 2022
[ Paper ]
LabOR: Labeling Only if Required for Domain Adaptive Semantic Segmentation
ICCV 2021 [Oral]
Received Qualcomm Innovation Award 2021
[ Paper ]
Discover, Hallucinate, and Adapt: Open Compound Domain Adaptation for Semantic Segmentation
NeurIPS 2020
Received Qualcomm Innovation Award 2021
[ Paper ]
Preserving Semantic and Temporal Consistency for Unpaired Video-to-Video Translation
MM 2019
[ Paper ]