Sanghyun Jo | OGQ · SNU AIBL Lab
Sanghyun Jo
Sanghyun Jo
Principal AI Researcher
shjo.april@gmail.com

Biography

I lead AI research and development at OGQ, where I have been building practical and interpretable machine learning systems since 2017. My primary focus is on designing models that prioritize tangible real-world impact , emphasizing robustness, transparency, and usability alongside strong empirical performance.

Since 2021, I have been conducting research under the supervision of Prof. Kyungsu Kim at the SNU AIBL Lab. This collaboration has deeply influenced my approach to explainable AI and data-efficient learning. Integrating these insights, I strive to develop AI systems that are not only powerful but also understandable and responsible.

-
Publications
-
Years in AI Industry

News

2026.02
TRACE selected as Oral at ICLR 2026 (223 papers, 1.13%) Oral
2026.01
TRACE accepted to ICLR 2026
2025.06
Two papers (COIN, ELECT) accepted to ICCV 2025
2025.06
Named Outstanding Reviewer for CVPR 2025 Award
2024.07
Two papers (DHR, TTD) accepted to ECCV 2024
2023.07
MARS accepted to ICCV 2023
2021.01
Puzzle-CAM accepted to ICIP 2021

Publications

Under ReviewEraseLoRA
EraseLoRA: MLLM-Driven Foreground Exclusion and Background Subtype Aggregation for Dataset-Free Object Removal
Sanghyun Jo*†, Donghwan Lee*, Eunji Jung*, Seong Je Oh, Kyungsu Kim†
arXiv:2512.21545
Co-first & Co-corresponding author
Under ReviewACE
On the Collapse of Generative Paths: A Criterion and Correction for Diffusion Steering
Ziseok Lee*, Minyeong Hwang*, Sanghyun Jo, Wooyeol Lee, Jihyung Ko, Young Bin Park, Jae-Mun Choi, Eunho Yang†, Kyungsu Kim†
arXiv:2512.10339
Under ReviewISAC
ISAC: Training-Free Instance-to-Semantic Attention Control for Improving Multi-Instance Generation
Sanghyun Jo*, Wooyeol Lee*, Ziseok Lee*, Kyungsu Kim†
arXiv:2505.20935
Co-first author
ICLR'26 OralTRACE
TRACE: Your Diffusion Model is Secretly an Instance Edge Detector
Sanghyun Jo*, Ziseok Lee*, Wooyeol Lee, Jonghyun Choi, Jaesik Park†, Kyungsu Kim†
International Conference on Learning Representations (ICLR), 2026.
Co-first author; Oral [Top 1.13%]
ICCV'25COIN
COIN: Confidence Score-Guided Distillation for Annotation-Free Cell Segmentation
Sanghyun Jo*, Seo Jin Lee*, Seungwoo Lee, Seohyung Hong, Hyungseok Seo†, Kyungsu Kim†
IEEE/CVF International Conference on Computer Vision (ICCV), 2025.
Co-first author
ICCV'25ELECT
Early Timestep Zero-Shot Candidate Selection for Instruction-Guided Image Editing
Joowon Kim*, Ziseok Lee*, Donghyeon Cho, Sanghyun Jo, Yeonsung Jung, Kyungsu Kim†, Eunho Yang†
IEEE/CVF International Conference on Computer Vision (ICCV), 2025.
ECCV'24DHR
DHR: Dual Features-Driven Hierarchical Rebalancing in Inter-and Intra-Class Regions for Weakly-Supervised Semantic Segmentation
Sanghyun Jo, Fei Pan, In-Jae Yu, Kyungsu Kim†
European Conference on Computer Vision (ECCV), 2024.
First author
ECCV'24TTD
TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate Single Tag Bias
Sanghyun Jo*, Soohyun Ryu*, Sungyub Kim, Eunho Yang, Kyungsu Kim†
European Conference on Computer Vision (ECCV), 2024.
Co-first author
ICCV'23MARS
MARS: Model-agnostic Biased Object Removal without Additional Supervision for Weakly-Supervised Semantic Segmentation
Sanghyun Jo, In-Jae Yu, Kyungsu Kim†
IEEE/CVF International Conference on Computer Vision (ICCV), 2023.
First author
PreprintRSEPM
RecurSeed and EdgePredictMix: Pseudo-Label Refinement Learning for Weakly Supervised Semantic Segmentation
Sanghyun Jo, In-Jae Yu, Kyungsu Kim†
arXiv:2204.06754
First author
ICIP'21PuzzleCAM
Puzzle-CAM: Improved localization via matching partial and full features
Sanghyun Jo, In-Jae Yu†
IEEE International Conference on Image Processing (ICIP), 2021.
First author; 150+ citations

Services & Awards

Academic Services
Conference Reviewing
CVPR 2025–2026
ICLR 2026
ECCV 2026
ICCV 2025
ICIP 2022–2024
ICPR 2024
ICML 2022
Journal Reviewing
IJCV 2024–2026
IEEE TCSVT 2022
Awards
Jun 2025
Selected as one of 711 outstanding reviewers out of 12,593 (Top 5.6%).
Jun 2023 – Jul 2023
1st out of 1,153 participants. Case representation and multi-branch modeling for predicting civil court rulings.
May 2023 – Jun 2023
5th out of 1,747 participants. Robust vehicle detector under domain shift.
Talks
ECCV 2024 Paper Presentation
MODULABS
Dec 2024
Watch
Advanced AI Developments
Yonsei University
Nov 2023
Strategies for Reducing Labeling Costs in Vision AI
Google Developer Groups (GDG)
Oct 2023
Experience Sharing Session: Vision Research
MODULABS
Nov 2022
Watch
RSEPM Paper Presentation
MODULABS
Oct 2022
Watch