This is adopted from https://github.com/rmsalinas/DBow3.
I do not lay any claims to its development. It is just a version of it which is easy
to use (atleast for me). Only the CMakeLists.txt file by me. This only depends on
OpenCV 3.
The publication to cite in case you use this work is:
@ARTICLE{GalvezTRO12, author={G'alvez-L'opez, Dorian and Tard'os, J. D.}, journal={IEEE Transactions on Robotics}, title={Bags of Binary Words for Fast Place Recognition in Image Sequences}, year={2012}, month={October}, volume={28}, number={5}, pages={1188--1197}, doi={10.1109/TRO.2012.2197158}, ISSN={1552-3098} } }
The core files all exist in folder dbow3_core. If you use this, you likely just need to #include "DBoW3.h".
In the repo's root directory you will find a) 3 usage-examples, b) a pretrained vocabulary (orbvoc.dbow3), c) a few sample images.
- create_voc_step0.cpp : Given a descriptor's name viz. brisk, SURF etc, and a folder containing ton of images. This can extract features and feature-descriptors for each of the images and save all of it to disk for later processing.
- create_voc_step1.cpp : Using the descriptors computed from step0, does clustering on the features to get make a vocabulary.
- demo_general.cpp : Loads pretrained vocabulary and a few images and queries them from the DB.
- dbow3_test.cpp : A more realistic usage case. This is what a SLAM's loop closure module would look like.
mkdir build
cmake ..
make
./get_orbvoc.sh
It is rather crude to link all the .o with the executables (see the CMakeLists.txt). I will accept a PR (pull-request) if you change this bit to make a static-linked library (.a) for the core files.
Optimized version of DBOW with AVX/SSE instructions https://github.com/rmsalinas/fbow
Manohar Kuse [email protected]
rmsalinas