Jaeyoo Park


Jaeyoo Park

Ph.D. Student

Computer Vision Lab
Department of Electrical and Computer Engineering
Seoul National University
email
github
google scholar
linkedin

About Me

I am a PhD student majoring in computer vision at SNU Computer Vision Lab, advised by Prof. Bohyung Han. My research focuses on unifying multiple modalities and tasks through shared multimodal representations, with particular interest in vision-language models, visual text and document understanding, and multimodal understanding and generation. This direction builds on my broader work on efficient learning under limited supervision, including continual, self-supervised, and weakly supervised learning in video and incremental learning settings. My recent work extends this perspective to unified multimodal models, with a focus on connecting understanding and generation pathways within a common framework.

News

  • [May. 2026] I am selected as a Gold Reviewer in ICML 2026.
  • [Oct. 2025] I am selected as a Top Reviewer in NeurIPS 2025.
  • [Jul. 2025] I joined Amazon AGI as a applied scientist intern.
  • [Apr. 2025] Our paper about text readability of vision-language models has been accepted to EVAL-FoMo 2 Workshop at CVPR 2025.
  • [Sept. 2024] Our paper about OCR-free document understanding has been accepted to NeurIPS 2024.
  • [June 2024] I joined Naver AI Lab as a research intern.
  • [June 2024] I am selected as an Outstanding Reviewer in CVPR 2024.

Work Experience

Education

Publications

→ Full list

Structural Self-Teaching for Compositional Generation in Unified Multimodal Models

Jaeyoo Park, Bohyung Han

Under Review

[Paper (Coming soon)]


DiVaTe: A Benchmark for Semantic Contamination in Visual Text Rendering

Yoonho Kim*, Jaeyoo Park*, Wonjae Roh, Bohyung Han

Under Review

[Paper (Coming soon)]

(* : equal contribution)


Emergence of Text Readability in Vision Language Models

Jaeyoo Park, Sanghyuk Chun, Wonjae Kim, Sangdoo Yun, Bohyung Han

In EVAL FoMo Workshop @ CVPR 2025

[Paper] [Poster]


Hierarchical Visual Feature Aggregation for OCR-Free Document Understanding

Jaeyoo Park, Jinyoung Choi, Jeonghyung Park, Bohyung Han

In NeurIPS, 2024

[Paper] [arXiv] [Poster] [bibtex]


Cross-Class Feature Augmentation for Class Incremental Learning

Taehoon Kim, Jaeyoo Park, Bohyung Han

In AAAI, 2024,

In CLVision Workshop @ CVPR 2024

[arXiv]


Multi-Modal Representation Learning with Text-Driven Soft Masks

Jaeyoo Park, Bohyung Han

In CVPR, 2023

[Paper] [Supp] [arXiv] [Poster] [bibtex]

Qualcomm Innovation Fellowship Korea 2023 Winner [Link]


End-to-End Learning for Weakly Supervised Video Anomaly Detection using Absorbing Markov Chain

Jaeyoo Park*, Junha Kim*, Bohyung Han

In CVIU (Impact Factor : 4.886), 2023

[Paper]

(* : equal contribution)


Class-Incremental Learning by Knowledge Distillation With Adaptive Feature Consolidation

Minsoo Kang, Jaeyoo Park, Bohyung Han

In CVPR, 2022 (Oral Presentation)

[Paper] [Supp] [arXiv] [Code] [bibtex]


Class-Incremental Learning for Action Recognition in Videos

Jaeyoo Park, Minsoo Kang, Bohyung Han

In ICCV, 2021

[Paper] [Supp] [arXiv] [Code] [Poster] [bibtex]


Rotation-Invariant Local-to-Global Representation Learning for 3D Point Cloud

Seohyun Kim, Jaeyoo Park, Bohyung Han

In NeurIPS, 2020

[Paper] [Supp] [arXiv] [Code] [bibtex]


Learning to Adapt to Unseen Abnormal Activities under Weak Supervision

Jaeyoo Park*, Junha Kim*, Bohyung Han

In ACCV, 2020

[Paper] [Supp] [arXiv] [Code] [bibtex]

(* : equal contribution)

Honors & Awards

  • Gold Reviewer, ICML 2026
  • Top Reviewer [Link], NeurIPS 2025
  • Outstanding Reviewer [Link], CVPR 2024
  • Qualcomm Innovative Fellowship Korea 2023 Winner [Link], Qualcomm Technologies, Inc., 2023
  • Youlchon AI Star Scholarship [Link], Youlchon Foundation & SNU AI Institute, 2023
  • Outstanding Reviewer [Link], ICCV 2021

Academic Services

  • Journal Reviewer
    • TPAMI, CVIU, MVAP

  • Conference Reviewer
    • CVPR, NeurIPS, ECCV, ICCV, ICLR, ICML, and IROS