Skip to content

geogismx/Broad-UNet

 
 

Repository files navigation

Broad-UNet: Multi-scale feature learning for nowcasting tasks

The official code of the following paper: https://arxiv.org/abs/2102.06442

Results

Some results for the different nowcasting tasks

Task Actual Vs Prediction
Precipitation prediction (30 mins ahead)
figures/ExampleRainPrediction20dataset-1.png
Precipitation prediction (30 mins ahead)
figures/ExampleRainPrediction50dataset-1.png
Cloud cover prediction (30 mins ahead)
figures/ExampleCloud30minsAhead-1.png
Cloud cover prediction (90 mins ahead)
figures/ExampleCloud90minsAhead-1.png

Installation

The required modules can be installed via:

pip install -r requirements.txt

Quick Start

Depending on the nowcasting task to be performed, the models can be trained running:

python training_clouds_data.py

or

python training_precipitation_data.py

To evaluate the models and visualize some predictions, please run:

python evaluation_and_predictions_clouds.py

or

python evaluation_and_predictions_precipitation.py

Scripts

  • The scripts contain the models, the generators, the training files and evaluation files.

Broad-UNet architecture

figures/Broad-UNet.PNG
figures/ConvBlock.PNG
figures/ASPP.PNG

Data and pretrained models

In order to download the data or any of the trained models, please email to the following addresses:

[email protected]

[email protected]

The data must be downloaded and unzipped inside the 'dataset_clouds/' or 'dataset_precipitation' directories as indicated in the txt files inside them.

Citation

If you use our data and code, please cite the paper using the following bibtex reference:

@misc{fernandez2021broadunet,
      title={Broad-UNet: Multi-scale feature learning for nowcasting tasks},
      author={Jesus Garcia Fernandez and Siamak Mehrkanoon},
      year={2021},
      eprint={2102.06442},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%