标题:Spatial pyramid formulation in weakly supervised manner
作者:Yue, Yawei ;Yue, Zhongtao ;Wang, Xiaolin ;Ni, Guangyun
作者机构:[Yue, Yawei ;Wang, Xiaolin ;Ni, Guangyun ] School of Computer Science and Technology, Shandong University, 250001 Jinan, China;[Yue, Zhongtao ] School 更多
会议名称:10th International Symposium on Neural Networks, ISNN 2013
会议日期:4 July 2013 through 6 July 2013
来源:Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
出版年:2013
卷:7952 LNCS
期:PART 2
页码:421-428
DOI:10.1007/978-3-642-39068-5-51
关键词:Local Sensitive Hash; Spatial Pyramid Match; Weakly Supervised
摘要:Spatial pyramid match scheme (SPM) is an important scheme in local feature based image classification which effectively adopts geometric structure information into image classification. Most previous approach formulized Spatial Pyramid in unsupervised manner by hierarchical splitting images into separate bins. We found that weak supervised information exists in this process totally unused. We cannot use information directly, because those information corresponding to the combination of all bins, thus we can use those weakly supervised information for bins selection. In this paper, we proposed to select those bins with better discriminative properties.The discriminative property can be well defined from neighborhood entropy. We incorporate local sensitive hash for fast neighborhood identification. We set those bins with higher neighborhood entropy weight zero. Analysis shows that our approach can down weight those non-discriminative bins, in contrast highlighting those discriminative bins. Experiments show that our approach can improve the performance of spatial pyramid match, especially for those categories with complex background. We also proof that under our scheme, result kernel matrix can still preserve positive semi-definite, which can guarantee that our algorithm will coverage. © 2013 Springer-Verlag Berlin Heidelberg.
收录类别:EI;SCOPUS
资源类型:会议论文;期刊论文
原文链接:https://www.scopus.com/inward/record.uri?eid=2-s2.0-84880707709&doi=10.1007%2f978-3-642-39068-5-51&partnerID=40&md5=25f3eac8a83d183a232e65ec86358725
TOP