Home My Page Projects Yael
Summary Activity Forums SCM Files

Project Filelist for Yael

File Release Notes and Changelog

Release Name: v1845

Release Notes
What is it?
===========

This is a Matlab package that we provide to reproduce the results of our CVPR paper (paper homepage: http://lear.inrialpes.fr/pubs/2010/JDSP10)

@InProceedings{JDSP10,
  author       = "Herv\'e J\'egou and Matthijs Douze and Cordelia Schmid and Patrick P\'erez",
  title        = "Aggregating local descriptors into a compact image representation",
  booktitle    = "IEEE Conference on Computer Vision \& Pattern Recognition",
  month        = "jun",
  year         = "2010",
  url          = "http://lear.inrialpes.fr/pubs/2010/JDSP10"
}

This package is outdated now. The webpage containing the improved version of the proposed image search system is available on the webpage http://lear.inrialpes.fr/src/inria_fisher/

Prerequisites
=============

The prerequisites are a working version of Matlab/mex. We have tested the software with relatively recent versions (R2009b, R2010a). We have made the package run with the open-source software Octave (3.2.3). The installation procedure differs a bit, but skilled Octave users should be able to make the mex-files of Yael work with octave.



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

These instructions are for Linux and MacOS X (just take care for
64-bits Matlab to use "long" instead of "int" for sgemm calls). This package is not supported for Windows. Sorry.

The commands below should be lanched from the directory where you unzipped this package.

1) Download the Yael library and the pqcodes package. 

The current package library has been tested with version SVN_v62 of the Yael library, but forward compatibility should be preserved. The Yael library can be obtained from the website: 
https://gforge.inria.fr/frs/?group_id=2151

The pqcodes package is available from the software webpage of Herve Jegou


2) Untar the lib and compile its Matlab interface (in subdirectory yael). 
> tar xvzf yael_v62.tar.gz
> rm yael_v62.tar.gz
> cd yael
> ./configure.sh
> cd matlab
> make
> cd ../..

If this does not work on your platform, please take a look at the README file and the Yael getting started manual. 


3) Compile additional Mex-file: 
> tar xvzf pqsearch_1830.tar.gz
> rm pqsearch_1830.tar.gz
> mex siftgeo_read_fast.c
> cd pqcodes
> mex sumidxtab.c
> cd ..


4) Get the SIFT descriptors associated with the Holidays database
> wget http://pascal.inrialpes.fr/data/holidays/siftgeo.tar.gz
> tar xvzf siftgeo.tar.gz
> rm siftgeo.tar.gz

5) Launch the test program in matlab:
>> test