EVENT

Event News

ASPIRE Human-Centered AI - UMCS Monthly Talk - Dr. Takuma Yagi (AIST)

How can AI understand human behavior and communication from a first-person perspective? In this month's UMCS Monthly Talk, we will explore egocentric vision and discuss its relevance to human-centered AI, sign language understanding, and multimodal communication research.

Title:

Egocentric Vision: Towards human-centered Computer Vision

Presenter:

Dr. Takuma Yagi (Senior Researcher, National Institute of Advanced Industrial Science and Technology (AIST))(Collaborator of UMCS)

Abstract:

Egocentric video refers to footage captured by wearable cameras mounted on mobile agents such as humans, vehicles, and robots. Because the camera records the environment from the agent's own viewpoint, it delivers rich, first‑person context that fixed third‑person cameras (e.g., surveillance cameras) cannot provide. This unique perspective is increasingly vital for applications ranging from AR assistants that support daily activities and work to autonomous robots that must understand and interact with their surroundings.
This talk introduces the foundations and current progress of egocentric video understanding, with a focus on how first-person visual data can support human-centered computer vision. I will review representative datasets such as Ego4D and Ego-Ego4D, and discuss recent approaches for (i) data collection and (ii) modeling fine-grained human activities. I will try to give relevance of egocentric vision to sign language analysis and communication understanding, where fine-grained motion, hand shape, body movement, gaze, and context are essential.

Bio:

Takuma received his Ph.D. degree from The University of Tokyo, Japan in 2022. He is currently a senior researcher at National Institute of Advanced Industrial Science and Technology (AIST), Japan. His research interests include Computer Vision and Machine Learning for human behavior understanding, especially in Egocentric Vision, Video Understanding, and Hand-Object Interaction Understanding.