Visual Odometry

---

Table of Contents

• Prerequisites
• Embedded Installation
• Developer Installation
• Usage

Prerequisites

First install Python 3.12 or greater. Then create a Python virtual environment and update pip if necessary. All dependencies will be installed to the virtual environment and will not affect your system install. If needed you can remake the virtual environment by deleting the venv folder.

python3.12 -m venv venv
. venv/bin/activate
pip install -U pip

You will need to activate the virtual environment in every new terminal.

. venv/bin/activate

Embedded Installation

You can install the package directly from the GitHub repository. Installing from ssh is recommended but you can use the https url instead if needed.

pip install "git+ssh://git@github.com/schebell-q/CS5330-Project.git"

or

pip install "git+https://github.com/schebell-q/CS5330-Project.git"

Developer Installation

Clone this repository and then use a developer mode install (editable install). Doing this allows pip to use the source code directly and you can edit it without reinstalling.

git clone git@github.com:schebell-q/CS5330-Project.git project
cd project
python3.12 -m venv venv
. venv/bin/activate
pip install -U pip
pip install -e .

Usage

Installing this module creates a vo command. Use --help for an explanation of the possible arguments.

vo --help

This is an example of using the SIFT feature detector with a 95% RANSAC threshold to estimate the trajectory of data/lab01_trim.MOV.

vo data/lab01_trim.MOV -r 0.95 -f sift -o lab01_trim.png

Test data

We included test data referenced in our report and presentation in the data/ folder. These files are stored with git-lfs and additional setup is needed to fully download and use them.

You may need to install git lfs to your system. Instructions can be found on their project page: https://git-lfs.com

Next, configure git to use git-lfs.

git lfs install