Skip to main content
Integrated M.S./Ph.D. Student, KAIST CVLAB

Seonghu Jeon.

I explore intelligence, aiming to connect agentic robotics, 3D/4D world modeling, and human-robot interaction. My work so far spans multi-view generation, geometric foundation models, and visuomotor policy learning.

Portrait of Seonghu Jeon

News & notes

    1. Education

      Moving to the integrated M.S./Ph.D. program at KAIST, starting Spring 2027.

      KAIST CVLAB, advised by Prof. Seungryong Kim. M.S. since March 2026.

      Experience

    2. Paper

      CSFM was accepted to NeurIPS 2026.

      Better Source, Better Flow: Learning Condition-Dependent Source Distribution for Flow Matching

      paperprojectcode

    3. Paper

      GAM was accepted to 3DWM @ ECCV 2026.

      Geometric Action Model for Robot Policy Learning

      paperprojectcode

    4. Honor

      GLD was selected for a Long Oral Presentation at ECCV 2026.

      Repurposing Geometric Foundation Models for Multi-view Diffusion

      paperprojectcode

    5. Career

      Joined MAUM AI's WoRV Division, Future Intelligence Lab as a research intern for Robotics Foundation Model development.

      Research Intern, 2026.07 – Present.

      Experience

    6. Education

      Joined KAIST CVLAB as an M.S. student under Prof. Seungryong Kim.

      Multi-view diffusion, geometric foundation models, 4D scene generation.

      Experience

    7. Paper

      CAMEO accepted to CVPR 2026 for correspondence-attention alignment in multi-view diffusion.

      CAMEO: Correspondence-Attention Alignment for Multi-View Diffusion Models

      paperprojectcode

    8. Education

      Graduated from Korea University, B.S. in Computer Science & Engineering, with Great Honors.

      Great Honors, GPA 4.46 / 4.50.

      Experience

    1. Paper

      ReMoTE was accepted to ITC-CSCC 2025 as an oral presentation.

      ReMoTE: A Benchmark for Object Motion Transfer. First author.

      paper

    2. Honor

      Won 2nd Prize at the IEIE Undergraduate AI Paper Competition (AISP 2025).

      Revisiting Flow-Conditioned Motion Transfer via Pseudo-Flow and Consecutive Local Attention

    1. Honor

      Received the National Science and Technology Scholarship from KOSAF.

    2. Career

      Moved to KAIST CVLAB with my supervisor, continuing as an undergraduate research intern.

    3. Community

      Started as DevRel Lead of GDGoC KU, Google Developer Groups on Campus at Korea University.

    4. Honor

      Received the merit-based Sangah Scholarship.

    5. Honor

      Received the merit-based Woonhae Scholarship.

    6. Community

      Started as President of AIKU, the Korea University AI Society.

    1. Career

      Joined CVLAB at Korea University as an undergraduate research intern under Prof. Seungryong Kim.

    2. Honor

      Won 1st Prize at the Korea-Yonsei University Datathon.

    3. Honor

      Won the Grand Prize at InThon, Korea University College of Informatics.

Featured research

All publications

Focus

Agentic Robotics

Prefrontal cortex: planning what to do and in what order.

Human-Robot Interaction

Language areas: communicating with people.

3D/4D World Modeling

Cerebellum: geometric understanding and world modeling, for predicting the future and precise manipulation.

Experience

  1. 2026.07 – Present

    Research Intern

    MAUM AI, WoRV Division, Future Intelligence Lab

    • Participating in Robotics Foundation Model development.
  2. 2026.03 – Present

    Integrated M.S./Ph.D. Student

    KAIST CVLAB

    • Advised by Prof. Seungryong Kim.
    • M.S. from March 2026, integrated M.S./Ph.D. program from Spring 2027.
    • Multi-view diffusion, geometric foundation models, 4D scene generation.
  3. 2023.12 – 2026.02

    Research Intern

    CVLAB, Korea University → KAIST

    • Researched generative 3D/4D vision, multi-view diffusion, flow matching, and geometric foundation models.
    • Built evaluation benchmarks and geometry-aware action models for motion transfer, visuomotor control, and embodied scene understanding.
    • Transferred to KAIST with the supervisor in Aug 2024.
  4. 2022 – 2026

    B.S., Computer Science & Engineering

    Korea University

    • Graduated with Great Honors (4.46 / 4.50).
    • Coursework in vision, graphics, and deep learning.