Profile

Seunghwan Lee

Researcher at ETRI.

Greetings,

I am a researcher at ETRI, working on efficient representation and compression of immersive visual content. Prior to this, I received my M.S. from Sungkyunkwan University, where I was advised by Prof. Sungeun Hong in the AI & Media Lab.

My research focuses on computer vision and machine learning, particularly model compression, efficient adaptation, and 3D scene representation techniques such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS).

Papers

Publications
  • MultihopSpatial
    MultihopSpatial: Multi-hop Compositional Spatial Reasoning Benchmark for Vision-Language Model
    Y Lee, S Jang, Y Cho, Seunghwan Lee, Y Lee, S Hwang · ECCV 2026
    Paper Homepage Code Dataset
  • SyMerge
    SyMerge: From Non-Interference to Synergistic Merging via Single-Layer Adaptation
    A Jung, Seunghwan Lee, D Han, S Hong · ICML 2026
    Paper Homepage Code
  • Task Vector Quantization
    Task Vector Quantization for Memory-Efficient Model Merging
    Y Kim*, Seunghwan Lee*, A Jung*, B Ryu, S Hong · ICCV 2025
    Paper Homepage Code
  • COTICA
    Instance-Aware Test-Time Segmentation for Continual Domain Shifts
    Seunghwan Lee*, I Jung*, H Lee, E Park, S Hong · arXiv preprint 2025
    Paper
  • PCA
    Prototypical Class-Wise Test-Time Adaptation
    H Lee, Seunghwan Lee*, I Jung*, S Hong · Pattern Recognition Letters 2025
    Paper

Background / Experience

Work Experience

2026–Present: Researcher at ETRI

2023: Research Intern at VCL Lab

2020: Consulting Intern at TG

Education

2024–2026: Sungkyunkwan University, M.S. Immersive Media Engineering

2017–2024: Inha University, B.E. Information and Communication Engineering

Awards

Feb. 2026: Honorable Mention for Outstanding Paper Award, IPIU 2026

May. 2025: 3rd place, Infrared Instance Segmentation Challenge, ICRA 2025

Feb. 2025: Honorable Mention for Outstanding Paper Award, IPIU 2025

Dec. 2023: Excellence Award, Capstone Design for Information and Communication Engineering, Inha University

Scholarships & Grants

2024–2025: Sensor Fusion and Missing Modality Handling for Occluded Instance Segmentation in Autonomous Driving — M.S. Students Fellowship by the National Research Foundation of Korea (NRF)