Junho Kim

Hi, I’m Junho

I am a Postdoctoral Researcher at the University of Illinois Urbana–Champaign, working with Prof. James M. Rehg on large-scale multimodal intelligence. I completed my Ph.D. at KAIST, where I was advised by Prof. Yong Man Ro.

My research aims to develop socially intelligent multimodal AI that can understand human behavior and interactions in dynamic, real-world environments. I study how models perceive and reason about visual, linguistic, and nonverbal signals, and how these signals evolve over time. Building on my work in reliable multimodal reasoning and dynamic video understanding, my long-term goal is to develop AI agents that can understand ongoing social interactions and proactively assist people in the physical world.

Research Interests

• Socially Intelligent Multimodal Agents: developing models that understand human behavior, nonverbal social cues, and interaction dynamics to enable socially aware AI agents and robots.
• Dynamic Video Understanding: understanding evolving events and temporal dynamics, enabling AI systems to perceive, reason, and respond continuously.
• Reliable Multimodal Reasoning: improving the robustness, grounding, and reasoning capabilities of large multimodal models across diverse tasks and real-world environments.

Work Experience

• University of Illinois Urbana-Champaign, USA (Sep. 2025 - Present)
  - Postdoctoral Research Associate
  - Advisor: Prof. James M. Rehg

Education

• KAIST, Daejeon, South Korea (2019 Aug. - 2025 Aug.)
  - Ph.D. in Electrical Engineering
  - Advisor: Prof. Yong Man Ro
  - Dissertation: Advancing Factuality in Large Vision-Language Models through Selective Visual Information and Textual Context Refinement
• Dongguk University, Seoul, South Korea (2012 Feb. - 2019 Feb.)
  - B.S. in Electrical and Electronic Engineering

News

Show more ▾

Selected Papers

* Co-First Authors, † Corresponding Authors

Honors & Awards

• NeurIPS 2025 Top Reviewer Award (Top 8%) [Link] - Conference on Neural Information Processing Systems (NeurIPS), 2025
• Korea Advanced Institute of Science and Technology (KAIST) Fellowship (2019 - 2025)
• The World Embedded software Contest [Link] - Finalist with AI six-legged robot ‘Homecoming’ (2017)

Services and Talk

Area Chair & Program Committee & Reviewers

• CVPR 2023,2024,2025,2026
• ICCV 2023,2025
• ECCV 2022,2024,2026
• NeurIPS 2023,2024,2025,2026
• AAAI 2025
• ICLR 2024,2025,2026, 2027
• WACV 2026
• ACL ARR 2026

Journal Reviewers

• IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
• IEEE Transactions on Circuits and Systems for Video Technology (TCSVT)
• IEEE Transactions on Audio, Speech, and Language Processing (TASLP)

Invited Talk

• Guest Lecture at CS 598 (CVH), University of Illinois Urbana–Champaign, From LLMs to VLMs and Reinforcement Learning, 2026
• Invted Talk at 2026 TechArt Conference, CAU, What Should Multimodal LLMs Look At in Video?: Selective Perception in Video Understanding, 2026
• Invited Talk at AMI LAB, KHU, Video Understanding with Large Vision-Language Models: Challenges and Progress, 2025
• Guest Lecture at CS 598 (CVH), University of Illinois Urbana–Champaign, Evolutions of LLMs and VLMs, 2025
• Invited Talk (세상과 만나는 강연) at Gimpo High School, Shaping the Future of AI: from History to Large Language Model, 2025
• Invited Talk at SAIL LAB, KAIST, Advancing Factuality in Large Multi-Modal Models through Integrated Cross-Modal Consistency, 2025