ABOUT

I’m a Ph.D. student at ACSL @ GIST AI, advised by Prof. Ue-Hwan Kim. My research lies at the intersection of computer vision and robotics, with a focus on learning effective 2D/3D scene representations for embodied perception and reasoning. I investigate how fine-grained visual perception can be transformed into structured, geometrically grounded, and semantically rich representations that support robotic reasoning, planning, and interaction. Ultimately, I aim to bridge perception and high-level reasoning, enabling robots to build task-relevant world models and operate robustly in complex real-world environments.

* Keywords: 2D/3D Scene Understanding, Robotic AI, Representation Learning

Publications

  • OP3DSG: Open-vocabulary Part-aware 3D Scene Graph Generation for Real-world Environments
    Yirum Kim, Ue-Hwan Kim
    European Conference on Computer Vision (ECCV) 2026 [Project][Paper][Code]

  • MA3DSG: Multi-Agent 3D Scene Graph Generation for Large-Scale Indoor Environments
    Yirum Kim, Jaewoo Kim, Ue-Hwan Kim
    IEEE International Conference on Robotics & Automation (ICRA) 2026 [Paper][Code]

Corresponding author, * Equal Contribution

Awards and Honors

Education

Work Experiences

Teach Experiences