PyDBoW2

This a python binding for the DBoW2 implementation in ORB-SLAM2. I use their original implementation so that you can easily benchmark their loop detection results.

Installation

Requirements

1. Run git submodule init && git submodule update to clone the pybind11 and DLib into the current repo.

2. This repo needs OpenCV. Set export OpenCV_DIR=/path/to/opencv/cmake/config. This path could be the build folder of the opencv.

3. Build DLib repo.

cd thirdparty/DLib
mkdir build && cd build && cmake .. && make

4. run python -m pip install -e .. You can find examples in test.py.

Build Issues

If you cannot compile DLib, one problem I find is that the include files is not correct for BRIEF256.h file. Add these lines in BRIEF256.h

#include <DUtils/DUtils.h>
#include <opencv2/core/core.hpp>
#include <opencv2/imgproc.hpp>

TODO

1. I notice that the ORB Feature in ORB-SLAM2 is different from the OpenCV implementation.

2. Add init.py

From ORB-SLAM2

You should have received this DBoW2 version along with ORB-SLAM2 (https://github.com/raulmur/ORB_SLAM2). See the original DBoW2 library at: https://github.com/dorian3d/DBoW2 All files included in this version are BSD, see LICENSE.txt

We also use Random.h, Random.cpp, Timestamp.pp and Timestamp.h from DLib/DUtils. See the original DLib library at: https://github.com/dorian3d/DLib All files included in this version are BSD, see LICENSE.txt