|||
http://www.vision.ee.ethz.ch/~yangme/
Meng Yang(杨猛博士后) 很多论文在稀疏字典学习方面,尤其是FDDL 等,可用在动作识别。不少文章在 的主页上有code.
论文阅读1
IJCV2014上 Sparse Representation Based Fisher Discrimination Dictionary Learning for Image Classification(FDDL)
摘要:
基于稀疏表示的图像识别中,字典起了很关键的作用。本文提出了一个新的字典学习(DL)的新方法来改进分类性能。基于Fisher判别准则的,一个结构化的字典,对应着类别标号的字典原子被不断地学习;这样稀疏编码后的重建误差用来模式分类,同时,Fisher判别准则也加到稀疏系数上,使得稀疏系数具有小的类内方差和大的类间方差。
基于重建误差和稀疏系数这两种信息同时用来分类,本文提出了一个Fisher判别式的字典学习机制(FDDL)。在实验方面,与基于稀疏表示和字典学习的机制做了对比。
在科学网上的主业为:
http://id.sciencenet.cn/u/mikemengyang 博 客:http://blog.sciencenet.cn/u/mikemengyang 邮 箱:yangmengpolyu@gmail.com;yang.meng@szu.edu.cn 所在 机构:Computer science & software engineering school, Shenzhen University 职务/职称:Associate Professor |
2009.9 - 2012.9 Department of Computing, Hong Kong Polytechnic University, Hong Kong, Ph.D in Computer Science
2006.9 - 2009.4 School of Automation, Northwestern Polytechnical University, Xi'an, Master of Control theory and application
2005.9 - 2006.6 School of Automation, Northwestern Polytechnical University, Xi'an, Bachlor of Information engineering
2002.9 - 2005.6 Honors College, Northwestern Polytechnical University, Xi'an
2012.11 - 2014.04 Computer vision lab, ETH Zurich, Switzerland, Postdoc Fellow with Prof. Luc Van Gool
2012.09 - 2012.10 Department of Computing, Hong Kong Polytechnic University, Hong Kong, Research Associate
International Conference
CVPR 2014
ECCV 2014
ACCV 2014(deadline 20 June)
My seminar reports:
Dictoinary learning for image classification
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-7-28 00:41
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社