THU-luvision / MILD

open source code for loop closure detection, binary feature based.

MILD: Multi-Index hashing for Loop closure Detection.

Official implementation for MILD: Multi-Index hashing for Loop closure Detection. Paper

This is a project from LuVision SIGMA, Tsinghua University. Visit our website for more interesting works: http://www.luvision.net/

License

This project is released under the GPLv3 license. We only allow free use for academic use. For commercial use, please contact us to negotiate a different license by: fanglu at tsinghua.edu.cn

Citing

If you find our code useful, please kindly cite our paper:

@INPROCEEDINGS{8019479,
  author={Han, Lei and Fang, Lu},
  booktitle={2017 IEEE International Conference on Multimedia and Expo (ICME)}, 
  title={MILD: Multi-index hashing for appearance based loop closure detection}, 
  year={2017},
  volume={},
  number={},
  pages={139-144},
  doi={10.1109/ICME.2017.8019479}}

Prerequisites

Ubuntu 14.04

cmake 3.2.0

OpenCV 3.1 http://xfloyd.net/blog/?p=987

eigen3

octave (optional, only used for evaluation)

Installation

$ mkdir build

$ cd build

$ cmake ..

$ make

Usage

./mild imagelist.txt settings.yaml

input:

imageList.txt: indicats the path of each input RGB image per line settings.yaml: indicats the parameters used in loop closure detection

output:

output/imagelist/lcd_shared_flag.bin: detected loop closure are set as 1. To be used in the run_scritp.m to check the accuracy of the detected loop closure. output/imagelist/lcd_shared_score_mild.bin: the image similarity calculated using MILD. output/imagelist/relocalization_time_per_frame.bin: lcd time of each frame.

evluation: (based on MATLAB/OCTAVE)

evaluation('build/output/imageList_NewCollege/lcd_shared_flag.bin','build/output/imageList_NewCollege/lcd_shared_probability.bin','data/truthNewCollege.mat');