By Yoones Rezaei, Stephen Lee
If you find our paper helpful in your work, please consider citing:
@misc{https://doi.org/10.48550/arxiv.2205.12464,
doi = {10.48550/ARXIV.2205.12464},
url = {https://arxiv.org/abs/2205.12464},
author = {Rezaei, Yoones and Lee, Stephen},
keywords = {Computer Vision and Pattern Recognition (cs.CV), Artificial Intelligence (cs.AI), Machine Learning (cs.LG), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {sat2pc: Estimating Point Cloud of Building Roofs from 2D Satellite Images},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
In this repository we release the code and data for our paper "sat2pc: Generating Building Roof's Point Cloud from a Single 2D Satellite Images" in ICCPS 2023. You can find the original paper here.
To use this repository you need an environemnt with python 3.7.9. We suggests creating a conda environment with the following command:
conda create -n "sat2pc" python=3.7.9
Next, activate the environemnt and cd to the direcotry of the downloaded repository then run the following command:
python create_conda_environemt.py
This command will install the required packages.
The dataset from the paper can be dowloaded from here.
To train the model you can run the following command:
python train.py --config ./configs/sat2pc.gin --data-dir ./datasets/
To test the model you can run the following command:
python test.py --config ./configs/sat2pc.gin --data-dir ./datasets/ --ckpt-path [location of the saved weights]
To visualize the predictions from the model you can use the following command:
python visualize.py --data-dir ./datasets --results [location of the .res file generated by running the test script]