Prof. Yu Wang of Hitotsubashi University, Japan

Video attribute recognition can be divided into objective attribute recognition and subjective attribute recognition. Objective attribute recognition is primarily concerned with recognizing tangible and stable characteristics, such as action or scene categories in a video. On the other hand, subjective attribute recognition is significantly more challenging since subjective attributes, such as violence extent or social relation atmosphere in a video, are defined by human perceptual cognition and are influenced by individual experiences or emotions. These attributes are not only difficult to define, but also difficult to recognize. In this talk, we introduce our recent works on ground-truth generation protocols, attribute recognition pipelines, and model explanation methods for subjective video attribute recognition. Our contributions consist of novel video datasets and training/explaining approaches.

Biography:

He received the M.S. degree in Information Science and the Ph.D. degree in Engineering, from Nagoya University, in 2010 and 2013, respectively. He was a JSPS research fellow from 2012 to 2013, a JSPS foreign research fellow from 2014 to 2015, and a designated assistant professor in 2016, with the Graduate School of Information Science, Nagoya University. He was an assistant professor with the Graduate School of International Development, Nagoya University, from 2017 to 2018. He was an assistant professor with the College of Information Science and Engineering, Ritsumeikan University, from 2019 to 2021. He is currently an associate professor with the Center for Information and Communication Technology, Hitotsubashi University. His research interests are image recognition and video analysis. He is a member of IEEE.

Address:Japan