Hyunjoon Lee

I am an integrated MS/PhD student at VGI Lab (Visual and Geometric Intelligence Lab) at Seoul National University, where I am advised by Prof. Jaesik Park.

Email  /  CV  /  Scholar  /  LinkedIn  /  Github

profile photo

Education

  • Seoul National University, Seoul, Korea
    Integrated M.S./Ph.D. in Interdisciplinary Program in Artificial Intelligence
    Advisor: Jaesik Park
    Mar. 2024 - Present
  • Sungkyunkwan University (SKKU), Suwon, Korea
    B.S. in Mechanical Engineering
    Summa Cum Laude(1/131), GPA: 4.39/4.5
    Mar. 2018 - Feb. 2024

Experiences

  • Research Intern, HuGe Lab, KAIST AI (Seoul, Korea)
    Advisor: Beomjoon Kim
    Jun. 2023 - Sep. 2023
    Developed a model to predict 6-DoF grasp poses using a depth sensor.
  • Research Intern, CSI Lab, SKKU (Suwon, Korea)
    Advisor: Yusung Kim
    Jan. 2023 - Jun. 2023
    Developed an RL agent using the PPO algorithm and transformer architecture.
  • Research Intern, CAMAS Lab, SKKU (Suwon, Korea)
    Advisor: Jachoon Koo
    Dec. 2019 - Mar. 2020
    Developed a mathematical model to elucidate the stiffness mechanism of the soft gripper.

Research

My research interests lie at the intersection of 3D vision and robotics. I am particularly interested in how we can leverage rich scene information to enhance robot manipulation capabilities in complex environments.

project image Hyunjoon Lee*, Eunsung Cha*, Jaesik Park,
Under Review, 2026

We propose a training-free hierarchical clustering framework for open-vocabulary 3D Gaussian segmentation, unifying instance-level coherence and query flexibility via an efficient digraph-based formulation without per-scene optimization.

project image Jinmo Kim, Namtae Kim, Hyunjoon Lee, Seungha Kim, Jaesik Park,
Under Review, 2026

We propose a feed-forward multi-view 3D reconstruction framework that explicitly reasons about occlusions by reconstructing missing latent tokens across views, enabling fast and consistent amodal shape completion without generative inpainting.

project image CF3: Compact and Fast 3D Feature Fields
Hyunjoon Lee, Joonkyu Min, Jaesik Park,
ICCV(Main Conference & Demonstrations Track), 2025
project page / arXiv / code

We propose a method to build compact and fast 3D Gaussian feature fields by effectively compressing and sparsifying Gaussians, achieving competitive performance with significantly fewer gaussians.

Awards & Honors

  • Honorable Mention, Research Paper Competition, Seoul National University, Korea (Dec. 2025)
  • Outstanding Paper Presentation Award, "Efficient Feature Lifting and Compression Using Pre-trained 3D Gaussians", Korea Computer Congress (Aug. 2025)
  • Invited Talk, "CF3: Compact and Fast 3D Feature Fields", SNU Graduate of AI Research Exchange (Jun. 2025)
  • Top Excellence Award at the Creative Research Project, Seoul National University (Jun. 2025)
  • Outstanding Award at the Demo Competition, Seoul National University (Jun. 2025)
  • Invited Talk, "CF3: Compact and Fast 3D Feature Fields", NaverLabs (Mar. 2025)
  • Dean's List Award, Sungkyunkwan University (May 2021)
  • Scholarship for Academic Excellence, Sungkyunkwan University (Mar. 2021)
  • Dean's List Award, Sungkyunkwan University (Oct. 2020)
  • Scholarship for Research Excellence, Sungkyunkwan University (Mar. 2020)
  • Scholarship for Academic Excellence, Sungkyunkwan University (Mar. 2019)
  • Scholarship for Academic Excellence, Sungkyunkwan University (Sep. 2018)

Patents

  • CF3: Compact and Fast 3D Feature Fields, KR Patent 10-2025-0093174

Template from Jon Barron's website.