/SVO

Stereo Visual Odometry with traditional algorithms

Primary LanguageC++MIT LicenseMIT

Stereo Visual Odometry

Stereo Visual Odometry Implementation in C++ on KIITI Dataset.

Requirements

  • OpenCV
  • Cmake

Usage

Manual
mkdir build
cd build
cmake ..
make
./output -h
Auto

Put the absolute path of KIITI Dataset in setup.sh file and run following.

sudo chmod +x setup.sh
./setup.sh

Incase of ERROR : header file not found edit .vscode/c_cpp_properties.json

Watch video

Image Calibration

To calibrate camera parameters take a picture of the $7*9$ chessboard from your camera and keep the ' Image.jpg ' in the data folder or specify the path and run the Helper.py file in src. It will save the data in parameters.txt in src folder. By default you can use the chessboard in data folder or you can specify the path of your own files.

Requirements

  • OpenCV
  • Numpy
  • Matplotlib
  • Argparser

For extra features run :

cd src
chmod +x Helper.py
./Helper.py -h

Dataset Directory tree

Mini Daatset Download KITTI Dataset Download Maintain this directory oder to avoid errors

.
├── image_00
│   ├── data  [108]
│   └── timestamps.txt
├── image_01
│   ├── data  [108]
│   └── timestamps.txt
├── image_02
│   ├── data  [108]
│   └── timestamps.txt
├── image_03
│   ├── data  [108]
│   └── timestamps.txt
├── oxts
│   ├── data  [108]
│   ├── dataformat.txt
│   └── timestamps.txt
└── velodyne_points
    ├── data  [108]
    ├── timestamps.txt
    ├── timestamps_end.txt
    └── timestamps_start.txt

Project Tree

.
├── CMakeLists.txt
├── LICENSE
├── README.md
├── Research
│   ├── 10.1007@s00530-020-00662-9.pdf
│   └── SVO Paper.pdf
├── include
│   ├── feature.h
│   ├── rigid.h
│   ├── stereo.h
│   ├── system.h
│   └── utils.h
├── requirements.txt
├── setup.sh
└── src
    ├── Helper.py
    ├── Parameters
    │   ├── LeftData.csv
    │   ├── RightData.csv
    │   ├── StereoData.csv
    │   └── Stereomap.xml
    ├── feature.cpp
    ├── main.cpp
    ├── stereo.cpp
    ├── system.cpp
    └── utils.cpp