Inseok Kong

prof_pic.jpg

I am an Undergraduate Student Researcher in the Robotics and Computer Vision Lab at University of Seoul (UOS). I am pursuing a double major in Geoinformatics and Artificial Intelligence.

Currently, my research focuses on 3D computer vision, with a particular emphasis on 3D point cloud processing and representation learning. I strive to pursue generalizable and robust research that can adapt to diverse real-world conditions.

Looking forward, I am eager to expand my studies into Vision-Language Models (VLMs), multimodal architectures, robotics Vision-Language-Action (VLA) models, and 3D reconstruction. I am deeply passionate about building intelligent algorithms that can seamlessly understand and interact with their physical environments.

If you are interested in my research projects or potential collaborations, please feel free to contact me.

News

Apr 2026 My first first-author paper, “MAMVI: 3D Test-Time Adaptation via Masked Multi-View Point Clouds”, has been accepted to ICPR 2026!
Feb 2026 My first co-authored paper, “APC: Transferable and Efficient Adversarial Point Counterattack for Robust 3D Point Cloud Recognition”, has been accepted to CVPR Findings 2026
Feb 2026 I have just launched my personal website.

Experience

July 2025 - Present Undergraduate Student Researcher at Robotics and Computer Vision Lab, University of Seoul
  • Advisor: Prof. Jiyoung Jung

Education

University of Seoul
  • B.S. in Geoinformatics (Mar 2022 - Present)
  • Double Major in Artificial Intelligence (Sep 2025 - Present)

Publications

  1. DIRECT: Distribution-aware Representation Regularization for 3D Test-Time Adaptation under Corruptions
    Geunyoung Jung,  Inseok Kong, and Jiyoung Jung
    Under Review 2026
    PDF Code
  2. MAMVI: 3D Test-Time Adaptation via Masked Multi-View Point Clouds
    Inseok Kong, Geunyoung Jung, and Jiyoung Jung
    International Conference on Pattern Recognition (ICPR) 2026
    PDF Code
  3. APC: Transferable and Efficient Adversarial Point Counterattack for Robust 3D Point Cloud Recognition
    Geunyoung Jung, Soohong Kim,  Inseok Kong, and Jiyoung Jung
    IEEE/CVF Conference on Computer Vision and Pattern Recognition Findings (CVPR Findings) 2026
    PDF Code

Honors and Awards

  • Grand Prize Award (Oct 2025)
    Seoul Metropolitan Government Big Data Campus Competition
    ◦ Conducted location analysis for autonomous village bus routes in Seoul.
  • Encouragement Award (Aug 2025)
    AI Convergence Problem Discovery Industry-Academia-Research Hackathon (AICOSS)
    ◦ Generated a storybook utilizing Diffusion models and LLMs.
  • Dean’s Award of the College of Engineering (Jan 2025)
    AICOSS Industry-Academia Collaborative Deep Learning Winter Bootcamp
    ◦ Performed image classification utilizing 2D domain Vision Transformer (ViT).