CHUNG-ANG UNIVERSITY
CHUNG-ANG UNIVERSITY
Building multimodal AI that understands, reasons, and remains reliable in the real world.
RESEARCH HIGHLIGHTS
At MULTI Lab, we explore how artificial intelligence can understand, reason about, and interact with the world through multiple modalities, guided by natural language. Our research spans complex multimodal understanding, efficient intelligence, trustworthy AI, and 3D spatial intelligence, with the goal of building intelligent systems that can be flexibly instructed, effectively reason, and reliably operate in the real world.
1. Advanced Multimodal Understanding
Advancing multimodal AI to understand and reason over complex visual information, including documents, charts, and other structured visual content.
VLMs · Visual Reasoning · Chart Understanding
2. Efficient Multimodal Intelligence
Making multimodal models more efficient, scalable, and accessible.
Token Pruning · Model Compression · Quantization
3. Robust & Trustworthy AI
Investigating model vulnerabilities and building reliable multimodal intelligence.
Adversarial Robustness · Model Reliability · Unlearning
4. 3D Vision & Spatial Intelligence
Reconstructing and understanding the 3D world through geometric and visual reasoning.
3D Reconstruction · Camera Pose Estimation · Visual Correspondence
NEWS
[2026-09] Our paper (ASQ) with LG AI Research accepted to NeurIPS 2026!
[2026-06] Our paper (AnchorPrune) led by our undergraduate student accepted to ECCV 2026!
[2026-06] Our paper (MMPL) accepted to MICCAI 2026! Congrats all coauthors!
[2026-05] Our paper (LASE) with LG AI Research accepted to ICML 2026! Congrats all coauthors!
[2025-08] Our paper (REFINE) led by our undergraduate student accepted to EMNLP 2025 (main)!
[2025-05] Our paper (NATR) accepted to KDD 2025! Congrats all coauthors!
[2025-02] Two papers accepted to CVPR 2025, one accepted as Highlight (Top 13.5% paper)! Congrats all coauthors!
[2024-12] Our paper (ImagePiece) accepted to AAAI 2025! Congrats all coauthors!
[2024-09] Bumsoo Kim (PI) joined Chung-Ang University, School of Computer Science & Engineering as an Assistant Professor!
[2024-07] Our paper (BiCA) was accepted to ECCV 2024 as Oral Presentation (Top 2.33% paper)! Congrats all coauthors!
Recruitment
We are looking for graduate/undergraduate students to research together in MULTI Lab at Department of CSE, Chung-Ang University. In our lab, we are interested in developing multi-modal AI algorithms associated with computer vision and natural language processing, continual learning for large multimodal models (LVLMs), AI security, embodied AI, 3D vision & robotics and medical AI. If you are interested in the above research fields, please contact me via e-mail (bumsoo@cau.ac.kr). Research internships for at least one semester are mandatory for students who are interested in joining our lab for graduate studies.
중앙대학교 소프트웨어학부 멀티모달 연구실에서 함께 연구할 대학원생 및 학부연구생을 모집합니다. 저희 멀티모달 연구실에서는 컴퓨터비전, 자연어처리 및 멀티모달 인공지능 알고리즘 개발, 초거대 멀티모달의 연속 학습 및 Embodied AI, 그리고 의료 인공지능 분야를 연구하고 있습니다. 위 연구분야에 관심있거나 진학 관련 상담을 원하는 학생은 제 메일(bumsoo@cau.ac.kr)을 통해 연락 바랍니다. 특히, 대학원 진학을 희망하는 경우 한 학기 이상 연구실 인턴은 필수적으로 요구됩니다.