skip to main content
Show Results with:

Combining local and global: Rich and robust feature pooling for visual recognition

Xiong, Wei et al.

Pattern recognition -- Elsevier Science -- Volume: 62 C; (pages 225-235) -- 2017

Online access

  • Title:
    Combining local and global: Rich and robust feature pooling for visual recognition
  • Author: Xiong, Wei;
    Zhang, Lefei;
    Du, Bo;
    Tao, Dacheng
  • Found In: Pattern recognition. Volume 62:Number C(2017); 201702; 225-235
  • Journal Title: Pattern recognition
  • Subjects: Pattern perception; Perception des structures Périodiques; Patroonherkenning; LCSH: Pattern perception; Dewey: 006.4
  • Rights: Licensed
  • Publication Details: Elsevier Science
  • Abstract: AbstractThe human visual system proves expert in discovering patterns in both global and local feature space. Can we design a similar way for unsupervised feature learning? In this paper, we propose a novel spatial pooling method within an unsupervised feature learning framework, namedRich and Robust Feature Pooling(R2FP), to better extract rich and robust representation from sparse feature maps learned from the raw data. Both local and global pooling strategies are further considered to instantiate such a method. The former selects the most representative features in the sub-region and summarizes the joint distribution of the selected features, while the latter is utilized to extract multiple resolutions of features and fuse the features with a feature balance kernel for rich representation. Extensive experiments on several image recognition tasks demonstrate the superiority of the proposed method.HighlightsA novel pooling method is proposed to extract features in an unsupervised framework.The proposed method learns features from both the global and the local feature space.Multiple resolutions of features are learnt.The proposed method is insensitive to the parameters and the input data.
  • Identifier: Journal ISSN: 0031-3203
  • Publication Date: 2017
  • Physical Description: Electronic
  • Shelfmark(s): ELD Digital store
  • UIN: ETOCvdc_100041885213.0x000001

Searching Remote Databases, Please Wait