Jun Hyung Kim
Korea Advanced Institute of Science and Technology · 情報科学
研究室紹介
Professor Jun Hyung Kim's research lab specializes in computational intelligence and multimedia data analysis, with a focus on image processing, stereo matching, and automated image co-registration for medical and real-world applications. The lab also investigates social media dynamics, particularly the detection and analysis of coordinated online behavior such as foreign-state troll activities on news platforms. Recent work extends into multimodal analysis of moral emotions in online video content, emphasizing cross-cultural comparisons and ethical implications of digital communication.
Research Overview
Research Output Trend
Figures are computed from collected data and may differ slightly.
Selected Papers
8Type 5 phosphodiesterase terminates the action of nitric oxide (NO) induced 3',5'-cyclic monophosphate (cGMP). Sildenafil inhibits this phosphodiesterase, increases cellular cGMP concentrations and enhances NO-induced smooth muscle relaxation. We investigated the effect of sildenafil on the oesophageal motor function of healthy subjects and patients with nutcracker oesophagus. Eight healthy volunteers and nine patients with nutcracker oesophagus participated in this study. The participants under
본 논문에서는 영상 분할(image segmentation)을 이용한 블록(block) 기반의 스테레오 정합(stereo matching) 기법을 제안한다. 블록 기반의 매칭 방식은 스테레오 정합 알고리즘 중 국부적 방법의 가장 대표적인 알고리즘이다. 그러나 스테레오 정합 문제에 적용될 때 가려진 영역(occlusion)이나 텍스쳐(texture) 정보가 부족한 영역에서 오정합 발생 확률이 매우 높아 만족할 만한 성능을 제공하지 못한다. 본 논문에서 제안하는 알고리즘은 영상을 분할하여 정확한 경계정보를 획득한 다음 한 영상의 작은 영역에 대하여 상관성이 가장 높은 영역이나 차이값이 가장 낮은 영역을 참조 영상에서 탐색하여 기존의 알고리즘보다 더 높은 정확도를 제공한다. Middlebury dataset에서 제공된 영상들을 활용하여 스테레오 영상에 적용한 실험 결과들은 제안된 알고리즘이 기존의 블록 기반 알고리즘보다 더 좋은 성능을 보여주는 것을 입증해준다.
Automated image co-registration is a process of matching an image to a reference image by deriving transformation parameters to shift co-incident points in the real world to match in the image space. In order to improve efficiency and effectiveness of the co-registration approach, the author proposed a pre-qualified area matching algorithm which is composed of feature extraction with Canny operator and area matching algorithm with cross correlation coefficient. For refining matching points, outl
This dataset contains 112,658,554 comments from 4,047,831 users across 4,267,824 articles on Naver News (2006–2025), including 24,068 suspected foreign-state troll users (70 known trolls and 23,998 detected trolls) and 4,023,763 non-troll users. Usernames are partially masked to protect user privacy. Appendix A provides a detailed summary of the dataset statistics. Title: Cross-National Information Attacks: A Two-Decade Analysis of Troll Behavior in Korea Paper link: TBD
Official artifact release accompanying the WebConf 2026 paper "Moral Outrage Shapes Commitments Beyond Attention: Multimodal Moral Emotions on YouTube in Korea and the US". This release includes: Fine-tuned multimodal moral emotion classification models Inference and evaluation scripts Documentation and usage examples This repository is archived via Zenodo to ensure long-term accessibility and reproducibility.
Official artifact release accompanying the WebConf 2026 paper "Moral Outrage Shapes Commitments Beyond Attention: Multimodal Moral Emotions on YouTube in Korea and the US". This release includes: Fine-tuned multimodal moral emotion classification models Inference and evaluation scripts Documentation and usage examples This repository is archived via Zenodo to ensure long-term accessibility and reproducibility.