[image 02883] 講演会のご案内（Feb 20 13:00-14:30, Osaka U., Prof. Jongwoo Lim)

[Yasuyuki Matsushita]

Image-MLの皆様

大阪大学の松下です。お世話になっております。

下記の通り、大阪大学吹田キャンパスにてJongwoo Lim先生をお迎えして講演会を開催いたします。
登録等必要ございませんので、ふるってご参加いただけましたら幸いです。

--

Time: 13:00-14:30, Feb 20, 2018
Place: A101, Graduate School of Information Science and Technology, Osaka University 
（大阪大学吹田キャンパス　情報科学研究科A棟 A101）

Speaker: Professor Jongwoo Lim, Hanyang University

Title: Tracking Persons-of-Interest / Hedged Deep Tracking

Abstract:
In this talk, I would like to introduce two recent works on visual tracking. One is off-line multi-target tracking in a video with multiple shots, which is also known as person re-identification. As people's appearance can drastically change in different shots, it is critical to model the individual faces for re-identification. We propose a deep-learning based network fine-tuning approach with novel symmetric triplet loss function. Another work is more traditional single object tracking, using deep learning features combined with parameter-free hedge algorithm. It combines the responses from multiple weak trackers using hedging algorithm to give robust tracking results.

Bio:
Jongwoo Lim is an associate professor at Dept. of Computer Science, Hanyang University, Seoul, Korea. Before joining Hanyang University in 2012, he was a software engineer at Google inc. from 2011, and a senior scientist at Honda Research Institute USA inc. from 2005. He received a Ph.D. degree in 2005 and an MS degree in 2003 from University of Illinois at Urbana-Champaign (UIUC), and a B.S. degree from Seoul National University, Korea in 1997. He has published more than 40 papers and served as area chairs or associate editors of prominent computer vision and robotics conferences and journals. His research interests include computer vision, robotics, and machine learning.


--------
Yasuyuki Matsushita
Graduate School of Information Science and Technology
Osaka University
yasumat ＠ ist.osaka-u.ac.jp