Spencer Armon (501508) and Josh Sapira (501443)
We trained a convolutional neural network to identify horizon lines in images. The windows branch serves as the main repository for this, which works well on windows operating systems, and the retrain-model-clean branch works on apple operating systems.
We got our data from Nathan Jacobs, co-author of Horizon Lines in the Wild. This paper is cited below
@article{DBLP:journals/corr/WorkmanZJ16,
author = {Scott Workman and
Menghua Zhai and
Nathan Jacobs},
title = {Horizon Lines in the Wild},
journal = {CoRR},
volume = {abs/1604.02129},
year = {2016},
url = {http://arxiv.org/abs/1604.02129},
eprinttype = {arXiv},
eprint = {1604.02129},
timestamp = {Mon, 13 Aug 2018 16:49:14 +0200},
biburl = {https://dblp.org/rec/journals/corr/WorkmanZJ16.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}
Our model was too large to push to GitHub, but it can be accessed below:
For Apple users:
https://drive.google.com/drive/folders/1xB2AqjdtzqTO5yOjzus88e4HOU2Y7NtS?usp=sharing For Windows users:
https://drive.google.com/file/d/1PFepqXJwjQXKck9LGJCLeOvqS1Aqm9on/view?usp=drive_link