Sangdoo Yun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Sangdoo Yun has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Young-Ho Kim
NAVER AI Lab, Seongnam, Republic of Korea
3 shared publications - Dongyoon Han
NAVER AI Lab
2 shared publications - Byeongho Heo
NAVER AI Lab
1 shared publication - DaEun Choi
School of Computing, KAIST, Daejeon, Republic of Korea
1 shared publication - Daeho Um
School of Electrical and Computer Engineering, University of Seoul, Seoul, South Korea
1 shared publication - Eunji Kim
Samsung Advanced Institute of Technology, Samsung Electronics Co. Ltd., 130 Samsung-ro, Yeongtong-gu, Suwon-si, Gyeonggi-do 16678, Republic of Korea
1 shared publication
Latest Publications
- Open Access
- 11 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 16 (2) , 1-34
De-Decay: Defusing Computer Vision Model Degradation through Scalable and Actionable Human-Data Alignment
Computer Vision (CV) models can become outdated after deployment as real-world data evolves, requiring intensive attention from AI engineers to address degraded performance through tasks like data relabeling to update models with new human perceptions. Interactive human-in-the-loop systems have considerable potential to enhance model-steering practices. However, such workflows reveal two challenges: (1) scalability, where labor demands increase with data size, and (2) actionability, where human insights do not readily transform into model revisions. Based on our formative study (S1) on the current challenges faced by CV professionals, we developed De-Decay, an end-to-end Human-Data Alignment system offering scalable label-less assessment and actionable insight transformation . This enables engineers to investigate degradation and auto-retrain models with AI support, such as image clustering and regeneration. Our summative study (S2) showed that De-Decay helped engineers effectively identify and address CV degradation. We discuss how future research can enhance scalability and actionability in AI evaluation systems for aligning AI behaviors with human mental models.
- 11 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 16 (2) , 1-34
Open AccessDe-Decay: Defusing Computer Vision Model Degradation through Scalable and Actionable Human-Data Alignment
Computer Vision (CV) models can become outdated after deployment as real-world data evolves, requiring intensive attention from AI engineers to address degraded performance through tasks like data relabeling to update models with new human perceptions. Interactive human-in-the-loop systems have considerable potential to enhance model-steering practices. However, such workflows reveal two challenges: (1) scalability, where labor demands increase with data size, and (2) actionability, where human insights do not readily transform into model revisions. Based on our formative study (S1) on the current challenges faced by CV professionals, we developed De-Decay, an end-to-end Human-Data Alignment system offering scalable label-less assessment and actionable insight transformation . This enables engineers to investigate degradation and auto-retrain models with AI support, such as image clustering and regeneration. Our summative study (S2) showed that De-Decay helped engineers effectively identify and address CV degradation. We discuss how future research can enhance scalability and actionability in AI evaluation systems for aligning AI behaviors with human mental models.
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-31
ClearFairy: Capturing Creative Workflows through Decision Structuring, In-Situ Questioning, and Rationale Inference
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-31
Open AccessClearFairy: Capturing Creative Workflows through Decision Structuring, In-Situ Questioning, and Rationale Inference
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-23
LingoQ: Bridging the Gap between EFL Learning and Work through AI-Generated Work-Related Quizzes
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-23
Open AccessLingoQ: Bridging the Gap between EFL Learning and Work through AI-Generated Work-Related Quizzes
- 12 May 2025
- Published by Springer Nature
- p. 71-91
RoCOCO: Robustness Benchmark of MS-COCO to Stress-Test Image-Text Matching Models
- 12 May 2025
- Published by Springer Nature
- p. 71-91
RoCOCO: Robustness Benchmark of MS-COCO to Stress-Test Image-Text Matching Models
- Top Cited Papers
- 3 November 2024
- Published by Springer Nature
- p. 289-305
Rotary Position Embedding for Vision Transformer
- 3 November 2024
- Published by Springer Nature
- p. 289-305
Top Cited PapersRotary Position Embedding for Vision Transformer
- Open Access
- 26 May 2022
- Published by Optica Publishing Group in Optics Express
- Vol. 30 (12) , 20927-20942
Optimization-based method removing highlights with flash/no-flash image
This paper proposes an optimization-based method to remove highlights from specular surfaces without any special device other than a flash, or any limited capturing environment required by existing methods. Accordingly, we formulate an optimization problem to estimate a highlight-free image from the difference image between a flash/no-flash image pair. This optimization will automatically find and recover erroneous pixels (saturated pixels and remaining highlights) in the difference image. The optimization is designed to recover erroneous pixels using the intensities of their neighboring pixels while retaining the other correct pixels to preserve the original image details. We evaluated the proposed method qualitatively and quantitatively through experimentation using datasets of synthetic and real-world scenes. In the experiments, the proposed method outperforms existing state-of-the-art methods.
- 26 May 2022
- Published by Optica Publishing Group in Optics Express
- Vol. 30 (12) , 20927-20942
Open AccessOptimization-based method removing highlights with flash/no-flash image
This paper proposes an optimization-based method to remove highlights from specular surfaces without any special device other than a flash, or any limited capturing environment required by existing methods. Accordingly, we formulate an optimization problem to estimate a highlight-free image from the difference image between a flash/no-flash image pair. This optimization will automatically find and recover erroneous pixels (saturated pixels and remaining highlights) in the difference image. The optimization is designed to recover erroneous pixels using the intensities of their neighboring pixels while retaining the other correct pixels to preserve the original image details. We evaluated the proposed method qualitatively and quantitatively through experimentation using datasets of synthetic and real-world scenes. In the experiments, the proposed method outperforms existing state-of-the-art methods.
- Open Access
Weakly Supervised Semantic Segmentation using Out-of-Distribution Data
Weakly supervised semantic segmentation (WSSS) methods are often built on pixel-level localization maps obtained from a classifier. However, training on class labels only, classifiers suffer from the spurious correlation between foreground and background cues (e.g. train and rail), fundamentally bounding the performance of WSSS. There have been previous endeavors to address this issue with additional supervision. We propose a novel source of information to distinguish foreground from the background: Out-of-Distribution (OoD) data, or images devoid of foreground object classes. In particular, we utilize the hard OoDs that the classifier is likely to make false-positive predictions. These samples typically carry key visual features on the background (e.g. rail) that the classifiers often confuse as foreground (e.g. train), so these cues let classifiers correctly suppress spurious background cues. Acquiring such hard OoDs does not require an extensive amount of annotation efforts; it only incurs a few additional image-level labeling costs on top of the original efforts to collect class labels. We propose a method, W-OoD, for utilizing the hard OoDs. W-OoD achieves state-of-the-art performance on Pascal VOC 2012.
Open AccessWeakly Supervised Semantic Segmentation using Out-of-Distribution Data
Weakly supervised semantic segmentation (WSSS) methods are often built on pixel-level localization maps obtained from a classifier. However, training on class labels only, classifiers suffer from the spurious correlation between foreground and background cues (e.g. train and rail), fundamentally bounding the performance of WSSS. There have been previous endeavors to address this issue with additional supervision. We propose a novel source of information to distinguish foreground from the background: Out-of-Distribution (OoD) data, or images devoid of foreground object classes. In particular, we utilize the hard OoDs that the classifier is likely to make false-positive predictions. These samples typically carry key visual features on the background (e.g. rail) that the classifiers often confuse as foreground (e.g. train), so these cues let classifiers correctly suppress spurious background cues. Acquiring such hard OoDs does not require an extensive amount of annotation efforts; it only incurs a few additional image-level labeling costs on top of the original efforts to collect class labels. We propose a method, W-OoD, for utilizing the hard OoDs. W-OoD achieves state-of-the-art performance on Pascal VOC 2012.
- 1 August 2021
- Published by Elsevier in Pattern Recognition
- Vol. 116
Region-based dropout with attention prior for weakly supervised object localization
- 1 August 2021
- Published by Elsevier in Pattern Recognition
- Vol. 116
Region-based dropout with attention prior for weakly supervised object localization
- 2 March 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 29 (6) , 2239-2252
Action-Driven Visual Object Tracking With Deep Reinforcement Learning
In this paper, we propose an efficient visual tracker, which directly captures a bounding box containing the target object in a video by means of sequential actions learned using deep neural networks. The proposed deep neural network to control tracking actions is pretrained using various training video sequences and fine-tuned during actual tracking for online adaptation to a change of target and background. The pretraining is done by utilizing deep reinforcement learning (RL) as well as supervised learning. The use of RL enables even partially labeled data to be successfully utilized for semisupervised learning. Through the evaluation of the object tracking benchmark data set, the proposed tracker is validated to achieve a competitive performance at three times the speed of existing deep network-based trackers. The fast version of the proposed method, which operates in real time on graphics processing unit, outperforms the state-of-the-art real-time trackers with an accuracy improvement of more than 8%.
- 2 March 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 29 (6) , 2239-2252
Action-Driven Visual Object Tracking With Deep Reinforcement Learning
In this paper, we propose an efficient visual tracker, which directly captures a bounding box containing the target object in a video by means of sequential actions learned using deep neural networks. The proposed deep neural network to control tracking actions is pretrained using various training video sequences and fine-tuned during actual tracking for online adaptation to a change of target and background. The pretraining is done by utilizing deep reinforcement learning (RL) as well as supervised learning. The use of RL enables even partially labeled data to be successfully utilized for semisupervised learning. Through the evaluation of the object tracking benchmark data set, the proposed tracker is validated to achieve a competitive performance at three times the speed of existing deep network-based trackers. The fast version of the proposed method, which operates in real time on graphics processing unit, outperforms the state-of-the-art real-time trackers with an accuracy improvement of more than 8%.