Kernel Codebooks for Scene Categorization

Abstract

This paper introduces a method for scene categorization by modeling ambiguity in the popular codebook approach. The codebook approach describes an image as a bag of discrete visual codewords, where the frequency distributions of these words are used for image categorization. There are two drawbacks to the traditional codebook model: codeword uncertainty and codeword plausibility. Both of these drawbacks stem from the hard assignment of visual features to a single codeword. We show that allowing a degree of ambiguity in assigning codewords improves categorization performance for three state-of-the-art datasets.

Cite

Text

van Gemert et al. "Kernel Codebooks for Scene Categorization." European Conference on Computer Vision, 2008. doi:10.1007/978-3-540-88690-7_52

Markdown

[van Gemert et al. "Kernel Codebooks for Scene Categorization." European Conference on Computer Vision, 2008.](https://mlanthology.org/eccv/2008/vangemert2008eccv-kernel/) doi:10.1007/978-3-540-88690-7_52

BibTeX

@inproceedings{vangemert2008eccv-kernel,
  title     = {{Kernel Codebooks for Scene Categorization}},
  author    = {van Gemert, Jan C. and Geusebroek, Jan-Mark and Veenman, Cor J. and Smeulders, Arnold W. M.},
  booktitle = {European Conference on Computer Vision},
  year      = {2008},
  pages     = {696-709},
  doi       = {10.1007/978-3-540-88690-7_52},
  url       = {https://mlanthology.org/eccv/2008/vangemert2008eccv-kernel/}
}