SecondCandidate

Introduction

Public pre-print of our paper at ICIP 2015

This project is the source code of the paper "Searching for nearest neighbors with a dense space partitioning". In this project, we try to create a library for nearest neighbors search with the following features:

An implementation of Product Quantization for Approximated Nearest Neighbor Search
An implementation of the inverted multi-index
An implementation of the method that is described in Searching for nearest neighbors with a dense space partitioning

If you use this source code, please also cite the following reference:

@inproceedings{Nguyen15, 
	author={Nguyen, T. and Matsui, Y. and Yamasaki, T. and Aizawa, K.}, 
	booktitle={Proc. ICIP}, 
	title={Searching for nearest neighbors with a dense space partitioning}, 
	year={2015}, 
	pages={1--1}, 
	organization = {IEEE},
}

References

[1] H. Jegou, M. Douze and C. Schmid, "Product quantization for nearest neighbor search," IEEE Trans. Pattern Anal. Mach. Intell., vol. 33, no. 1, pp. 117–128, 2011.

[2] A. Babenko and V. Lempitsky, "The inverted multi-index," IEEE Trans. Pattern Anal. Mach. Intell., vol. 37, no. 99, pp. 1247–1260, 2014.

[3] T. A. Nguyen, Y. Matsui, T. Yamasaki and K. Aizawa, "Searching for nearest neighbors with a dense space partitioning," in Proc. ICIP, IEEE, 2015.

Installation

Prerequisites

An Unix Operating System. Tested on Mac OS X 10.9 and Ubuntu 14.04.
CMake 2.8 or newer. For UNIX users, check your CMake version in terminal by cmake -version.
An GNU C++ Compiler or LLVM compiler that support C++ 11 and OpenMP.

Build

We use CMake as the build system. On terminal,

$ git submodule update --init
$ cmake -H. -Bbuild && cmake --build build -- -j4
# OR IF YOU WANT TO BUILD TESTS
$ cmake -DBUILD_TEST=ON -H. -Bbuild && cmake --build build -- -j4

This script will create binaries in your bin/ and lib/ directories.

Documentation

We are using Doxygen to generate the documentation. You can find that the style of comments for every methods that are implemented in these source codes are matching with Doxygen style. To generate the documentation:

$ cd ./doc
$ doxygen config.doxygen

Doxygen will create the documentation files in HTML and LaTeX format. You can also find samples and documentation in the Wiki of this project.

To developers

This project was built under Eclipse CDT so we leave the configuration files .cproject and .project of the project here. If you are using Eclipse, just import the folder that contains the source code.

marker68/second-candidate