Kernelized Local Descriptors
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

KDE - Kernel Local Descriptors

This is a Matlab package that implements the kernel local descriptors presented in "Kernel Local Descriptors with Implicit Rotation Matching" at ICMR 2015

What is it?

This code implements:

  1. Extraction of our local descriptor
  2. Descriptor post-processing (power-law and PCA dimensionality reduction)
  3. Evaluation on the Brown patch dataset


Run the following script:

>> extract
>> evaluate


False Positive Rate at 95% recall on Brown patch dataset:

Train set Test set RootSIFT (128d) RootSIFT-PCA (80d) KDE_{3,2,2}-PCA (80d)
notredame liberty 29.64 19.34 11.73
yosemite liberty 29.64 19.95 12.42
notredame yosemite 26.69 18.37 7.17
liberty yosemite 26.69 19.52 8.39
liberty notredame 22.06 13.90 5.09
yosemite notredame 22.06 13.98 5.19


If you use this work please cite our publication "Kernel Local Descriptors with Implicit Rotation Matching" from ICMR 2015

  title     = {Kernel Local Descriptors with Implicit Rotation Matching},
  author    = {Bursuc, Andrei and Tolias, Giorgos and J{\'e}gou, Herv{\'e}},
  booktitle = {ACM International Conference on Multimedia Retrieval},
  year      = {2015},
  url       = {}


KDE is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see