2026-1학기 전산전자공학대학원 네 번째 세미나 강의를 소개합니다. We are pleased to announce the commencement of the 2026 Spring Semester Seminar Series for the Graduate School of Computer Science and Electrical Engineering. This weekly forum serves as a cornerstone for academic exchange, bringing together students, faculty, and industry experts to explore the frontiers of modern technology.
강사(Speaker): 김동근 교수 (한동대학교)
제목(Title): From Understanding Human Videos to Socially Intelligent Robots
일시(Date): 4월 29일 수요일, 오전 11시 30분 | April 29, Wednesday
장소(Venue): 뉴턴홀 220호
발표언어(Language): 한국어
강의초록(Abstract): This talk presents a series of research work on understanding collective human behaviors in videos, spanning group activity recognition, group activity detection, and fine-grained social interaction detection. Understanding human collective behavior requires compositional reasoning over multiple levels - from individual body-part cues such as gaze, gesture, and facial expression, up to pairwise interactions and group-level dynamics. Ultimately, such understanding forms the foundation for building socially intelligent systems that can perceive and reason about human activities in the real world. Looking ahead, I will discuss two future directions. The first is multimodal social activity understanding, extending beyond visual cues to incorporate language and audio for richer scene interpretation. The second is learning from human videos for robotic manipulation, which leverages large-scale human data to teach robots dexterous skills, ultimately moving toward multi-agent collaboration in complex environments.
본 세미나는 온라인으로 진행됩니다. 학부생, 대학원생, 교직원 누구나 참여 가능하니 많은 관심과 참여 바랍니다. This seminar will be conducted online. We welcome all undergraduate students, graduate students, faculty, and staff to join us. We look forward to your interest and active participation.
2026-1학기 전산전자공학대학원 네 번째 세미나 강의를 소개합니다. We are pleased to announce the commencement of the 2026 Spring Semester Seminar Series for the Graduate School of Computer Science and Electrical Engineering. This weekly forum serves as a cornerstone for academic exchange, bringing together students, faculty, and industry experts to explore the frontiers of modern technology.
Looking ahead, I will discuss two future directions. The first is multimodal social activity understanding, extending beyond visual cues to incorporate language and audio for richer scene interpretation. The second is learning from human videos for robotic manipulation, which leverages large-scale human data to teach robots dexterous skills, ultimately moving toward multi-agent collaboration in complex environments.
문의 (Inquiry): 054-260-3150(BK21 AI사업단)
본 세미나는 온라인으로 진행됩니다. 학부생, 대학원생, 교직원 누구나 참여 가능하니 많은 관심과 참여 바랍니다. This seminar will be conducted online. We welcome all undergraduate students, graduate students, faculty, and staff to join us. We look forward to your interest and active participation.