Kyungjin Lee

　

Seoul National University

Kyungjin Lee is a 4th-year PhD student in the Department of Computer Science and Engineering at Seoul National University, working under the supervision of Prof. Youngki Lee. Her primary research area is mobile computing, with intersections in computer vision and graphics, wireless networks, and human-computer interaction. She has authored numerous papers that have been presented at top-tier international conferences in computer science, including ACM MobiCom, IEEE INFOCOM, ACM CHI, and ACM SIGGRAPH. Additionally, she serves as a reviewer for UbiComp and ACM Computing Surveys, both top-tier conferences and journals in the field. Her research accomplishments have been acknowledged through multiple academic and industry research fellowships. She has also mentored more than 10 students, resulting in high-quality research outcomes, and has actively led student-organized academic events.

Everyday Hyper-realistic Telepresence Platform

My research vision is to develop a fully mobile end-to-end platform for hyper-realistic telepresence. Hyper-realistic telepresence eliminates the boundary between physical and virtual realities. It involves: i) creating a realistic digital representation of the real world through precise capture and three-dimensional (3D) reconstruction, ii) sharing the representation with multiple users, and iii) providing seamless and immersive interactions.

Making this platform accessible to everyday users could revolutionize the way we live and open up new possibilities. For example, long-distance couples and families can continuously feel as if they are living together, eating, chatting, and going on walks.

Major industry players like Google (Project Starline), Meta Research (Codec Avatars), and Microsoft (Holoportation) have introduced experimental systems highlighting the platform's potential. However, these systems require multiple high-quality cameras for real-time user capture, along with intricate offline scanning and model training, which are not scalable. Additionally, real-time reconstruction and data streaming demand powerful computers and wired data transfer.

In my research, I focus on building mobile software systems to realize such a platform. Today, we have numerous opportunities, including mobile devices equipped with high-quality sensors, advanced computer graphics and vision algorithms, especially deep neural networks (DNNs), commercially available immersive rendering devices, and widely available wireless networks. Yet, there still remains a significant gap between the system's goal and what current mobile systems can achieve.

My research aims to bridge this gap, demonstrating the feasibility of a fully mobile hyper-realistic telepresence platform, and identifying new areas for further innovation in related fields.