20200108-01 行人重识别Person Re-ID

报告时间：2020年01月08日（星期三）晚上21:00（北京时间）

主题：行人重识别Person Re-ID

报告主持人：魏云超（悉尼科技大学）

报告嘉宾：郑良（Australian National University）

报告题目：Thoughts about Object Re-identification and Beyond

报告嘉宾：常虹（中科院计算所）

报告题目：Feature representation in person Re-identification

Panel议题：

1. 在计算机视觉领域中，如何定位目标重识别这个研究问题呢？其与一些基本问题的区别和联系在哪呢？例如分类、检索等。

2. 如何解决重识别中存在的一些挑战性问题？比如同一个人穿不同的衣服，人体定位不准确，标注数据量少、有噪声，domain shift等问题。

3. 重识别的领域自适应问题和已有的领域自适应的区别和联系在哪呢？

4. 行人重识别与车辆重识别有什么区别和联系呢？

5. 目标重识别在多目标跟踪中的应用效果如何呢？

6. 未来研究方向。

Panel嘉宾：

郑良（Australian National University）、常虹（中科院计算所）、张史梁（北京大学）、郑伟诗（中山大学）、李鸿升（香港中文大学）

*欢迎大家在下方留言提出主题相关问题，主持人和panel嘉宾会从中选择若干热度高的问题加入panel议题！

报告嘉宾：郑良（Australian National University）

报告时间：2020年01月08日（星期三）晚上21:00（北京时间）

报告题目：Thoughts about Object Re-identification and Beyond

报告人简介：

Dr Liang Zheng is a Lecturer in the Research School of Computer Science, Australian National University. He obtained both his B.S degree (2010) and Ph.D degree (2015) from Tsinghua University. He makes some early attempts in large-scale person re-identification, and his works are positively received by the community. Dr Zheng received the Outstanding PhD Thesis and the Wen-Tsun Wu Award from Chinese Association of Artificial Intelligence, and the DECRA award from the Australian Research Council. His research is featured by MIT Technical Review, and some are selected into the computer science courses in Stanford University and University of Texas at Austin. He serves as an Area Chair/Senior PC in ECCV 2020, AAAI 2020, IJCAI 2019, and IJCAI 2020, and organizes tutorials and workshops at ECCV 2018, CVPR 2019 and CVPR 2020. He is an Associate Editor of IEEE TCSVT.

个人主页：

http://www.liangzheng.com.cn/

报告摘要：

The re-identification problem has been studied extensively studied in the past few years, and performance on some public datasets is close to saturation. In this talk, I will discuss some new perspectives that might be useful for the community. First, I will present our work connecting re-identification and multi-object tracking, through discussing the underlying differences between the two tasks. Second, I will discuss the use of synthetic data in re-identification and its potential applications in the broader computer vision community.

参考文献：

[1] Yunzhong Hou, Liang Zheng, Zhongdao Wang, Shengjin Wang. Locality aware appearance metric for multi-target multi-camera tracking. Arxiv 2019.

[2] Zhongdao Wang, Liang Zheng, Yixuan Liu, Shengjin Wang, Towards real-time multi-object tracking. Arxiv 2019.

[3] Xiaoxiao Sun, Liang Zheng, Dissecting person re-identification from the viewpoint of viewpoint. CVPR 2019.

[4] Yue Yao, Liang Zheng, Xiaodong Yang, Milind Naphade, Tom Gedeon, Simulating Content Consistent Vehicle Datasets with Attribute Descent. Arxiv 2019.

报告嘉宾：常虹（中科院计算所）

报告时间：2020年01月08日（星期三）晚上21:30（北京时间）

报告题目：Feature representation in person Re-identification

报告人简介：

Hong CHANG is currently a researcher in Institute of Computing Technology, Chinese Academy of Sciences. She received her Bachelor degree from Hebei University of Technology in 1998, M.Phil. degree from Tianjin University in 2001, and Ph.D. degree from Hong Kong University of Science and Technology in 2006, all in Computer Science. She was a permanent research scientist at Xerox Research Centre Europe. From 2008, she has been an associate researcher and researcher at Institute of Computing Technology, Chinese Academy of Sciences. Her main research interests include models and algorithms of machine learning and pattern recognition, especially semi-supervised learning, metric learning, few-shot learning, deep learning, etc., as well as the applications of machine learning methods in computer vision and pattern recognition, especially image and video modeling, object detection, tracking and person re-identification.

个人主页：

http://vipl.ict.ac.cn/people/~hchang

报告摘要：

Feature representation is a key problem in person re-identification. In this talk, I will introduce our recent works in person re-identification from the perspective of feature representation. Specifically, we try to learn person features with high robustness and discriminative ability, as well as with low information loss. Some discussions and future works will be mentioned at the end.

参考文献：

[1] Ruibing Hou, Hong Chang, Bingpeng Ma, Shiguang Shan, Xilin Chen, "Cross Attention Network for Few-shot Classification, "The Thirty-third Annual Conference on Neural Information Processing Systems (NeurIPS), 2019.

[2] Xinqian Gu, Bingpeng Ma, Hong Chang, Shiguang Shan, Xilin Chen, "Temporal Knowledge Propagation for Image-to-Video Person Re-identification," IEEE International Conference on Computer Vision (ICCV), 2019.

[3] Ruibing Hou, Bingpeng Ma, Hong Chang, Xinqian Gu, Shiguang Shan, Xilin Chen, “Interaction-and-Aggregation Network for Person Re-identification,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019.

[4] Ruibing Hou, Bingpeng Ma, Hong Chang, Xinqian Gu, Shiguang Shan, Xilin Chen, “VRSTC: Occlusion-Free Video Person Re-Identification,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019.

Panel嘉宾：张史梁（北京大学）

嘉宾简介：

张史梁，北京大学研究员，博士生导师。于2012年在中国科学院计算技术研究所获得博士学位，2012-2015年先后在美国德克萨斯州大学圣安东尼奥分校、NEC美国硅谷实验室从事研究工作。研究方向为海量多媒体信息检索与计算机视觉，专注于行人重识别、物体细粒度识别、场景理解研究。以第一/通讯作者在IEEE T-PAMI、TIP、TMM等权威国际期刊以及ICCV、CVPR、ACM MM、AAAI等权威国际会议发表论文50余篇。中组部“青年千人”计划入选者，首批北京市杰出青年科学基金获得者。获2016年教育部技术发明一等奖，中国计算机学会优秀博士学位论文、中科院优秀博士学位论文、微软学者奖等。先后主持国家自然科学基金面上、重大研发计划培育项目、国家重点研发计划等项目。

个人主页：

https://www.pkuvmc.com/

Panel嘉宾：郑伟诗（中山大学）

嘉宾简介：

郑伟诗博士，中山大学数据科学与计算机学院教授，机器智能与先进计算教育部重点实验室副主任。他主要面向大规模智能视频监控里的行人身份识别与动作分析，展开视频图像信息与信号的识别与预测研究，并围绕该应用开展大规模机器学习的算法和理论研究。关于面向大规模监控网络下的行人追踪问题，他在国内外较早和持续深入开展跨视域行人重识别的研究，发表一系列以跨视域度量学习为主线的研究工作，并最近集中展开无监督和弱标注学习建模，力图解决“大数据小标注”下的图像视频分析问题。他已发表120余篇主要学术论文，含12篇IEEE T-PAMI和IJCV论文和其他80余篇发表在其他图像识别和模式分类IEEE TIP、IEEE TNN、PR、IEEE TCSVT、IEEE TSMC-B等国际主流权威期刊和ICCV、CVPR、AAAI、IJCAI等计算机学会推荐A类国际学术会议。担任Pattern Recognition等期刊的编委，担任AVSS 2012、ICPR 2018、IJCAI 2019/2020、AAAI 2020、BMVC 2018/2019 Area Chair/SPC等。他是IEEE MSA TC 委员。他主持国家重点研发课题一项、国家自然科学基金委-广东大数据科学中心 中心项目(集成项目)课题一项及其他5个国家级项目。获广东省自然科学奖一等奖、广东省科学技术进步奖二等奖等；获国家优秀青年科学基金、英国皇家学会牛顿高级学者基金和广东省创新领军人才项目支持。

个人主页：

http://www.isee-ai.cn/~zhwshi/

Panel嘉宾：李鴻升（香港中文大学）

嘉宾简介：

Hongsheng Li is currently an assistant professor in the Multimedia Laboratory at the Chinese University of Hong Kong. He was an associate professor at University of Electronic Science and Technology of China between 2013-2015. He received the bachelor’s degree from East China University of Science and Technology in 2006, and the doctorate degree from Lehigh University, United States in 2012. He has published over 60 papers in top computer vision and machine learning conferences, including CVPR, ICCV, ECCV, NeurlPS. He won the Object Detection from Videos (VID) track of ImageNet challenge 2016 as the team leader and 2015 as a team co-leader. He was guest editors of International Journal of Computer Vision and Neurocomputing. His research interests include computer vision, machine learning, and medical image analysis.

个人主页：

https://www.ee.cuhk.edu.hk/~hsli/

主持人：魏云超（University of Technology Sydney）

主持人简介：

Yunchao Wei is an Assistant Professor in the Centre for Artificial Intelligence, University of Technology Sydney. He received his PhD degree from Beijing Jiaotong University in 2016. He has published over 40 papers in the top-tier journals/conferences. He received the 1st prize of the science and technology award from China Society of Image and Graphics in 2019, Discovery Early Career Researcher Award of Australian Research Council in 2018, 1st place of all human parsing tracks in the 2nd LIP Challenge, Excellent Doctoral Dissertation Awards of Chinese Institute of Electronics (CIE) in 2016, and 1st place of object detection task in ILSVRC 2014. He organizes the workshop on Learning from Imperfect Data (LID) in CVPR 2019, 2020, Real-World Recognition from Low-Quality Images and Videos (RLQ) in ICCV 2019, Look Into Person (LIP) in CVPR 2019. His research interests mainly lie in applying machine learning techniques to tackle computer vision problems such as object detection and semantic segmentation.

个人主页：

https://weiyc.github.io/

20-01期VALSE在线学术报告参与方式：

长按或扫描下方二维码，关注“VALSE”微信公众号（valse_wechat），后台回复“01期”，获取直播地址。

特别鸣谢本次Webinar主要组织者：

主办AC：魏云超（The University of Technology Sydney）

协办AC：欧阳万里（The University of Sydney）

责任AC：韩琥（中科院计算所）

VALSE Webinar改版说明：

自2019年1月起，VALSE Webinar改革活动形式，由过去每次一个讲者的方式改为两种可能的形式：

1）Webinar专题研讨：每次活动有一个研讨主题，先邀请两位主题相关的优秀讲者做专题报告（每人30分钟），随后邀请额外的2~3位嘉宾共同就研讨主题进行讨论（30分钟）。

2）Webinar特邀报告：每次活动邀请一位资深专家主讲，就其在自己熟悉领域的科研工作进行系统深入的介绍，报告时间50分钟，主持人与主讲人互动10分钟，自由问答10分钟。

活动参与方式：

1、VALSE Webinar活动依托在线直播平台进行，活动时讲者会上传PPT或共享屏幕，听众可以看到Slides，听到讲者的语音，并通过聊天功能与讲者交互；

2、为参加活动，请关注VALSE微信公众号：valse_wechat 或加入VALSE QQ群（目前A、B、C、D、E、F、G、H、I、J、K群已满，除讲者等嘉宾外，只能申请加入VALSE L群，群号：641069169）；

*注：申请加入VALSE QQ群时需验证姓名、单位和身份，缺一不可。入群后，请实名，姓名身份单位。身份：学校及科研单位人员T；企业研发I；博士D；硕士M。

3、在活动开始前5分钟左右，讲者会开启直播，听众点击直播链接即可参加活动，支持安装Windows系统的电脑、MAC电脑、手机等设备；

4、活动过程中，请不要说无关话语，以免影响活动正常进行；

5、活动过程中，如出现听不到或看不到视频等问题，建议退出再重新进入，一般都能解决问题；

6、建议务必在速度较快的网络上参加活动，优先采用有线网络连接；

7、VALSE微信公众号会在每周四发布下一周Webinar报告的通知及直播链接。

8、Webinar报告的PPT（经讲者允许后），会在VALSE官网每期报告通知的最下方更新[slides]。

9、Webinar报告的视频（经讲者允许后），会更新在VALSE爱奇艺空间，请在爱奇艺关注Valse Webinar进行观看。

郑良[slides]

常虹[slides]