Learning and Exploring Visual Similarity

17 March 2015
March 17, 2015

Time: 10:00
Location: Meeting Room Ofek, Polo Scientifico e Tecnologico "Fabio Ferrari" (Building Povo 1), via Sommarive 5 - Povo, Trento

Speaker:

  • Ying Wu
    Northwestern University, USA

Abstract:
Visual similarity is one of the most fundamental problems in computer vision and pattern recognition. It is a basis for almost all computer vision tasks, such as target tracking, image super-resolution, image search, object recognition, etc. Using pre-defined visual similarity metric is limited and inappropriate, because visual similarity is not only task-dependent, but also data-dependent. Therefore, a good visual similarity needs to adaptive and needs to be learned from data. In this talk, I will present our recent work on learning and exploring visual similarity, based on several case studies including target tracking and image super-resolution. One specific task is to match and track low-resolution targets in video, which are generally imaged in tens of pixels, and the fine visual details are lost.
This is a practical yet quite challenging task, as the lack of discriminative visual details leads to matching ambiguity. Performing image super-resolution as the pre-processing step is not plausible, because it is computationally demanding. In this talk, I will present a novel solution to this task without explicitly performing video super resolution. Instead, our new approach aims to learn a discriminative metric for low resolution matching.
This learned metric preserves the data affinity structure in the high resolution feature space. We call this approach discriminative metric preservation. Bypassing explicit video super resolution, the proposed approach results in effective and efficient matching for low resolution images.

About the Speaker:
Professor Ying Wu received the B.S. from Huazhong University of Science and Technology, Wuhan, China, in 1994, the M.S.
from Tsinghua University, Beijing, China, in 1997, and the Ph.D. in electrical and computer engineering from the University of Illinois at Urbana-Champaign (UIUC), Urbana, Illinois, in 2001. He is currently full professor of Electrical Engineering and Computer Science at Northwestern University, Evanston, Illinois. He joined Northwestern University as assistant professor in 2001, and was promoted to associate professor in 2007 and to full professor in 2012. His research interests include computer vision, image and video analysis, pattern recognition, machine learning, multimedia data mining, and human-computer interaction. He serves as associate editors for IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), IEEE Transactions on Image Processing (T-IP), IEEE Transactions on Circuit Systems and Video Technology (T-CSVT), SPIE Journal of Electronic Imaging (JEI), and IAPR Journal of Machine Vision and Applications (MVA). He served many times as area chairs for CVPR, ICCV, ICIP and ACM Multimedia. He will be Program Co-Chair of CVPR'17. He received the Robert T. Chien Award at UIUC in 2001, and the National Science Foundation (NSF) CAREER award in 2003. He is a senior member of the IEEE.

Contact Person Regarding this Talk (name and email): Nicu Sebe, niculae.sebe [at] unitn.it