Giyeol Kim

giyeol.jpg

I am currently a Masterโ€™s student in the Perceptual AI Lab at Chung-Ang University, advised by Prof. Chanho Eom. My research focuses on person search, multi-modal retrieval, and diffusion-based generative models. I am broadly interested in building learning systems that can understand complex relationships across heterogeneous modalities, and in developing scalable methods for robust representation learning. If youโ€™re interested in my work or would like to discuss potential collaborations, feel free to reach out.

Publications

  1. Preprint
    ECOVAD: Efficient Open-world Video Anomaly Detection
    Juhyun Park, Giyeol Kim, and Chanho Eom
    In arXiv, 2025
  2. Preprint
    DiCo: Disentangled Concept Representation for Text-to-image Person Re-identification
    Giyeol Kim and Chanho Eom
    In arXiv, 2025
  3. ICCV
    Leveraging Prior Knowledge of Diffusion Model for Person Search
    Giyeol Kim*, Sooyoung Yang*, Jihyong Oh, Myungjoo Kang, and Chanho Eom
    In IEEE/CVF International Conference on Computer Vision (ICCV), 2025
    ๐Ÿ† Highlight Paper (263/11152 = Top 2.35%)
  4. ICCVW
    Chest X-ray Feature Pyramid Sum Model with Diseased Area Data Augmentation Method
    Changhyun Kim, Giyeol Kim, Sooyoung Yang, Hyunsu Kim, Sangyool Lee, and Hansu Cho
    In Computer Vision for Automated Diagnosis (CVAMD) Workshop at IEEE/CVF International Conference on Computer Vision (ICCV), 2023

Reviewer Experiences

ECCV (2024), ICCV (2025), NeurIPS (2025)

Patents

10-2025-0133759 Sep. 17, 2025 (Application)
[KOR] ํ™•์‚ฐ ๋ชจ๋ธ์„ ์ด์šฉํ•œ ๊ฐ์ฒด๋ฅผ ๊ฒ€์ƒ‰ํ•˜๋Š” ๋ฐฉ๋ฒ• ๋ฐ ์ด๋ฅผ ์‹คํ–‰ํ•˜๊ธฐ ์œ„ํ•œ ์ปดํ“จํŒ… ์žฅ์น˜
[ENG] Diffusion Model-Based Multi-task Decoupling Method for Improving Person Search Accuracy
์—„์ฐฌํ˜ธ, ๊น€๊ธฐ์—ด, ์˜ค์ง€ํ˜•, ์–‘์ˆ˜์˜, ๊ฐ•๋ช…์ฃผ
์ค‘์•™๋Œ€ํ•™๊ต ์‚ฐํ•™ํ˜‘๋ ฅ๋‹จ