BAMapping

This framework can reconstruct 3D scenes from RGB-D images.

Author: Yi Zongyao

21 May 2020: Formal version

1.Prerequisites

C++11 Compiler

OpenCV

OpenCV is used to facilitate feature extraction and matching. Tested with openCV 3.4. Install instructions can be found at: http://opencv.org. Install the library under: "libs/opencv_libs/"

Open3d

We used Open3D to realize volumetric fusion and mesh visualization. Please refer to http://www.open3d.org. Install the library under: "libs/open3d_lib/"

Pangolin

The Pangolin is used for sparse point cloud and trajectory visualization. Install the library under: "libs/pangolin_lib/"

#pangolin:
#dependency:
#OpenGL
sudo apt install libgl1-mesa-dev
#Glew
sudo apt install libglew-dev
#CMake
sudo apt install cmake
#Wayland

#pkg-config: 
sudo apt install pkg-config
#Wayland and EGL:
sudo apt install libegl1-mesa-dev libwayland-dev libxkbcommon-dev wayland-protocols

cd Pangolin
mkdir build
cd build
cmake ..
cmake -DCMAKE_INSTALL_PREFIX="../libs/pangolin_lib" --build .
make install

DBoW2

The DBoW2 library is utilized to detect loop closures. Install the library under: "libs/dbow_lib/"

ORB vocabulary

For loop closure detection, we used the vocabulary file provided in ORB SLAM2. Download and extract then put it under "Voc/"

Ceres Solver

We used Ceres Solver to solve non-linear optimization problem. Install instruction can be found at: http://ceres-solver.org/installation.html. Install the library under: "libs/ceres_libs/"

2.Build

cd BAMapping/
mkdir build/
cd ./build/
cmake ..
cmake --build .

3.Deployment

Make sure that the vocabulary file ORBvoc.txt is under "Voc/" and adjust the parameters in configuration files! In TUM dataset, camera intrinsics are not always the same.

ITE dataset

The dataset folders should be organized as:

ITE_dataset/
├── depth/                   # depth images
├── ir/                      # infra red images
├── rgb/                     # color images
├── temp/                    # folder to store temporary files 
├── cameras.txt              # initial trajectory
├── ITE.yaml                 # configuration file
└── ITE_frontend.yaml        # configuration file for front end    

launch the front end

cd BAMapping/build/
runFrontendITE ../dataset/some_ITE_dataset/

launch the back end

runBackenITE ../dataset/some_ITE_dataset/

TUM dataset

First associate the dataset according to https://vision.in.tum.de/data/datasets/rgbd-dataset/tools and copy the generated file to dataset's path. The dataset folders should be organized as:

TUM_dataset/
├── depth/                   # depth images
├── rgb/                     # color images
├── temp/                    # folder to store temporary files 
├── ass.txt                  # association file 
├── TUM.yaml                 # configuration file
└── TUM_frontend.yaml        # configuration file for front end  

launch the front end

cd BAMapping/build/
runFrontendTUM ../dataset/some_ITE_dataset/

launch the back end

runBackenTUM ../dataset/some_ITE_dataset/

performace fr2desk fr1desk
fr1room fr3long_office