Dr. Soopil Kim | Computer Science | Best Researcher Award

Daegu Gyeongbuk Institute of Science and Technology | South Korea

Author Profile

Scopus

Orcid

Early Academic Pursuits 🎓

Dr. Soopil Kim's academic journey began with a Bachelor of Engineering in Robotics and Mechatronics Engineering from Daegu Gyeongbuk Institute of Science & Technology (DGIST), where he graduated Cum Laude. He continued his studies at DGIST, pursuing a Master’s and Ph.D. in the same field, focusing on medical image segmentation. His research during these years emphasized label-efficient segmentation models and limited pixel-level annotation, laying a strong foundation for his future work in deep learning and computer vision.

Professional Endeavors 💼

Dr. Kim's career has seen significant milestones, including a role as a Visiting Student at Stanford University's CNSLAB under the supervision of Prof. Kilian M. Pohl and Ehsan Adeli. Currently, he is a Post-Doctoral Research Fellow at the Medical Image & Signal Processing Lab (MISPL) at DGIST, where he works under Prof. Sang Hyun Park. His professional trajectory reflects a commitment to advancing the field of computer vision through innovative research and collaboration.

Contributions and Research Focus 🔬

Dr. Kim’s research is at the forefront of deep learning and computer vision. His work addresses the challenges of image segmentation with partially labeled datasets by developing federated learning strategies and few-shot segmentation techniques. His notable contributions include the creation of a medical image segmentation model that integrates meta-learning and bi-directional recurrent neural networks, a semi-supervised segmentation model based on uncertainty estimation, and a transductive segmentation model for industrial imaging. These advancements aim to improve the efficiency and accuracy of image segmentation processes.

Accolades and Recognition 🏆

Dr. Kim has received several awards that highlight his exceptional contributions to the field. Notably, he was ranked 3rd among 40 teams in the SNUH Sleep AI Challenge in 2021 and was honored with the Outstanding Student Award from the Department of Robotics and Mechatronics Engineering at DGIST in 2022. In 2024, he was recognized at the KCCV Oral/Poster Presentation Doctoral Colloquium for his work on label-efficient segmentation models.

Impact and Influence 🌍

Dr. Kim's research has made a significant impact on the field of computer vision, particularly in the area of image segmentation. His innovative approaches to handling partially labeled datasets and federated learning have the potential to advance both academic research and practical applications in medical imaging and beyond. His work on few-shot learning and uncertainty-aware models addresses critical challenges in the field, contributing to more robust and adaptable segmentation solutions.

Legacy and Future Contributions 🚀

As Dr. Kim continues his research, his focus on improving segmentation models and developing new methodologies promises to shape the future of computer vision. His commitment to exploring federated learning and few-shot learning techniques will likely drive further innovations in the field, offering solutions to complex challenges and enhancing the accuracy of image analysis across various applications.

 

Publications 📘


📄Few-shot anomaly detection using positive unlabeled learning with cycle consistency and co-occurrence features
Authors: Sion An, Soopil Kim, Philip Chikontwe, Jiwook Jung, Hyejeong Jeon, Jaehong Kim, Sang Hyun Park
Journal: Expert Systems with Applications
Year: 2024


📄Federated learning with knowledge distillation for multi-organ segmentation with partially labeled datasets
Authors: Soopil Kim, Haejun Park, Myeongju Kang, Kilian M. Pohl, Sang Hyun Park
Journal: Medical Image Analysis
Year: 2024


📄FedNN: Federated learning on concept drift data using weight and adaptive group normalizations
Authors: Myeongju Kang, Soopil Kim, Kwang-Hyun Jin, Kilian M. Pohl, Sang Hyun Park
Journal: Pattern Recognition
Year: 2024


📄Few Shot Part Segmentation Reveals Compositional Logic for Industrial Anomaly Detection
Authors: Soopil Kim, Sion An, Philip Chikontwe, Kilian M. Pohl, Sang Hyun Park
Conference: Proceedings of the AAAI Conference on Artificial Intelligence
Year: 2024


📄Uncertainty-aware semi-supervised few shot segmentation
Authors: Soopil Kim, Philip Chikontwe, Sion An, Sang Hyun Park
Journal: Pattern Recognition
Year: 2023


 

Soopil Kim | Computer Science | Best Researcher Award