Self-supervised Behavioral Imaging
▣ 제목(Title) : Self-supervised Behavioral Imaging
▣ 연사(Speaker) : Prof. Hyun Soo Park (University of Minnesota)
▣ 초청(Host) : Prof. Jaesik Park
▣ 일시(Date &Time) : 2022.9.16(Fri), 10 am
▣ 장소 (Venue) : Science Building Ⅱ Room 102 (제2공학관 102호)
▣ 언어(Language) : 영어 (English)
Zoom ID : 939 0792 0120 PW : 215698
Title: Self-supervised Behavioral Imaging
Humans transmit social signals through a number of nonverbal cues, including gaze direction, facial expression, and body gesture. These cues are often subtle, e.g., microscopic facial muscle movements of cynical smiles, very difficult for AI to read. This poses a critical challenge in collecting, annotating, and learning from data. In this talk, I will present a few instances of our recent effort that addresses this challenge through self-supervised learning. This includes leveraging multiview geometry, temporal coherence, and semantic prior. To the end, I will discuss our new commercialization effort that pushes behavioral understanding in the wild.
Hyun Soo Park is an Associate Professor at the Department of Computer Science and Engineering, the University of Minnesota (UMN), and CEO of Playtag. He is interested in computer vision approaches for behavioral imaging. He has received NSF’s CRII, NSF’s CAREER Awards, and CVPR 2021 Best Paper Honorable Mention Award. Prior to UMN, he was a Postdoctoral Fellow in GRASP Lab at University of Pennsylvania. He earned his Ph.D. from Carnegie Mellon University.