This repository contains an implementation of the line-based SfM pipeline proposed in the paper "Structure from Motion with Line Segments Under Relaxed Endpoint Constraints" by Branislav Micusik & Horst Wildenauer.
Link to the paper: Click here
Please note that this repository does not include all of the steps mentioned in the paper and some of the steps have also been modified.
All necessary data and codes are inside the src
directory.
conda create -n vis3d python=3.7.10
pip install -r requirements.txt
There are the three following examples:
main.py
: It implements the general pipeline to perform line-based SfM on natural images.main_toy_example1.py
andmain_toy_example2.py
: To test our pipeline, we created these two toy examples.
To test on your data, follow the template of main.py
If you find this implementation useful, please cite us as:
@software{bryan_german_pantoja_rosero_2021_4946844,
author = {Bryan German Pantoja-Rosero and Amir Rezaie},
title = {Amuya Line-based SfM},
month = jun,
year = 2021,
publisher = {Zenodo},
version = {v0.0},
doi = {10.5281/zenodo.4946844},
url = {https://doi.org/10.5281/zenodo.4946844}
}