- 연사: Prof. Eungjoo Lee (Assistant Professor, The University of Arizona)

- Title: Deep Learning-based (Medical) Image Analysis on Resource-Constrained Systems

- Abstract: Recent advances in deep learning have significantly improved the performance of computer vision systems across domains ranging from autonomous navigation to medical image analysis. However, realizing these capabilities in real-time, resource-constrained environments, including surgical systems, wearable health devices, and embedded platforms, remains a challenge. In this seminar, I will discuss how AI integrates perception, decision, and action through efficient and robust visual intelligence. The presentation will focus on two directions: (1) data-centric learning for model robustness, including the use of self-supervised and synthetic data generation techniques to overcome the limitations of labeled datasets, and (2) computationally efficient model development for practical deployment, highlighting lightweight network architectures and knowledge distillation for embedded and edge devices. By integrating these directions, I aim to advance multimodal perception and decision-making systems that bridge algorithmic innovation and translational impact, moving toward scalable and human-centric intelligence in both autonomous and medical environments.

- 발표언어: 영어

- 장소: IT1호관 313호

- 일시: 2025. 12. 23.(화) 16:00

- 초청교수: 정희철 교수