Home My Page Projects Yael
Summary Activity Forums SCM Files

Project Filelist for Yael

File Release Notes and Changelog

Release Name: v2188

Release Notes
What is this?
=============

A (slow) matlab implementation of the product quantizer search method. This library allows the search in a large vector dataset indexed by a limited amount of memory. It is an implementation of the method described in 

@Article{JDS11,
  author       = "Herv\'e J\'egou and Matthijs Douze and Cordelia Schmid",
  title        = "Product quantization for nearest neighbor search",
  journal      = "IEEE Transactions on Pattern Analysis \& Machine Intelligence",
  number       = "1",
  volume       = "33",
  pages        = "117--128",
  month        = "jan",
  year         = "2011",
  url          = "http://lear.inrialpes.fr/pubs/2011/JDS11"
}


Legal & Contact
===============

This package was written by Herve Jegou
Copyright (C) INRIA 2009-2011
Last change: February 2011. 
Contact: herve.jegou@inria.fr

This software is governed by the CeCILL license under French law and abiding by the rules of distribution of free software. 
See http://www.cecill.info/licences.en.html

INRIA might also provide an optimized C/python implementation of this library, either for research of commercial purpose. Do not hesitate to contact us. 


Installation
============

1) Download the yael library and compile the matlab function (matlab subdirectory). 
See https://gforge.inria.fr/projects/yael/

The pqcodes package was tested with version Yael v138. 

Note this library is maintained only for Linux (likewise MacOS). Do not ask for a windows implementation, we will not provide it. 

2) Put the subdirectory "matlab" of Matlab in the matlab path (done by the "addpath" Matlab command)

3) Compile the mexfile for better efficiency:
>> mex sumidxtab.c

4) Optionnally, download the siftsmall/sift/gist evaluation datasets for tests on real data

5) run the pq_test or the ivfpq_test sample programs

Remark: by default, the program uses an automatically generated dataset ('random'). You can download the same datasets as the one we used in our paper [JDS11] at the following address: http://corpus-texmex.irisa.fr
In that case, you have to adjust the "basedir" variable in the file pq_test_load_Vectors.m to fit your local path.