报告题目：Person Re-identification in the Wild

报告人：Liang Zheng, University of Texas at San Antonio

时间：2016年4月18日（周一）下午2:00-3:30

地点：北京大学计算机科学技术研究所1层106会议室

摘要：Automated entry and retail systems at theme parks, passenger flow monitoring at airports, behavior analysis for automated driving and surveillance are a few applications where detection and recognition of persons across a camera network can provide critical insights. This talk thus focuses on such task named person re-identification. Previous studies mainly 1) consider only the recognition part and 2) have limitations in data scale and richness. In the attempt to initiate research on real-life person re-identification, this talk will introduce three high-quality large-scale datasets, provide extensive benchmarking results and propose the associated scalable approaches. Progressively, contributions will be described from three aspects: 1) large-scale image-based person re-identification, 2) large-scale tracklet-based person re-identification, and 3) an end-to-end person re-identification system integrating both pedestrian detection and recognition. This talk will present several observations that are very different from previous works, and most importantly, point out future research directions made available by our works.

简历：Dr. Liang Zheng received his B.E. degrees in Life Science from Tsinghua University in 2010, and his PhD degree in Electronic Engineering from Tsinghua University in 2015. After that, he became a postdoc researcher in the University of Texas at San Antonio, working with Prof. Qi Tian. During his PhD study, he was a recipient of Travel Grant Award of CVPR 2015 and ICCV 2015, and one of his works was featured by MIT Technology Review as "The best of the rest from the Physics arXiv preprint server". He was a twice recipient of the "Promising Young Researcher" of the EE department in Tsinghua University. His research interest consists in large-scale person re-identification, image retrieval, classification, and deep learning. He has published 13 first-author papers, including 5 in CVPR/ICCV, 3 in IEEE Transaction on Image Processing/Multimedia and 1 paper in IJCV.