報告題目: Person Identification across Multiple Moving-Camera Videos
Abstract: The use of multiple moving cameras, such as various wearable cameras, provide a new perspective for video surveillance by simultaneously collecting videos from different and time-varying view angles. These videos can better cover the targets and scene of interest. For integrated analysis of such videos, it is important to relate the targets, especially the persons, across these videos and this can be very challenging given their different and time-varying view angles. In this talk, I will describe this new problem of cross-video person identification, discuss its difference from the traditional person re-identification, and then introduce the machine-learning based approaches that can extract view-invariant appearance, motion, and human pose features for handling this cross-video person identification problem.
Bio: Song Wang（王松） received the B.E. degree from Tsinghua University in 1994 and the Ph.D. degree in electrical and computer engineering from the University of Illinois at Urbana–Champaign in 2002. He is currently a professor in the Department of Computer Science and Engineering at University of South Carolina. His research interests include computer vision, image processing, and machine learning. He has published more than 110 papers in relevant journals and conferences, including IEEE-TPAMI, IJCV, IEEE-TIP, ICCV, CVPR, NIPS, AAAI and IJCAI, with more than 3000 Google Scholar citations. He is serving as an associate editor of IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE-TPAMI), Pattern Recognition Letters, and Electronics Letters. He is also serving as the Publicity/Web Portal Chair of the Technical Committee of Pattern Analysis and Machine Intelligence of the IEEE Computer Society.