Skip to content

Akatuoro/icon-gan

Folders and files

NameName
Last commit message
Last commit date

Latest commit

3daf969 · Apr 30, 2024

History

96 Commits
Apr 30, 2024
Jul 17, 2020
Sep 28, 2020
Apr 30, 2024
Apr 27, 2024
Apr 23, 2024
Apr 23, 2024
Oct 16, 2018
Jul 10, 2019
Jul 17, 2020
Oct 15, 2018
Apr 23, 2024
Apr 27, 2024
Apr 27, 2024
Apr 27, 2024
Oct 15, 2018
Apr 23, 2024
Jul 17, 2020

Repository files navigation

  icon Icon Generator

How to use:

Open IconGenerator.ipynb with Jupyter Notebook.

Tensorflow and Keras should be installed, as well as a range of standard libraries like numpy. The tensorflow docker container already contains all the necessary libraries.

Training Dataset

The easiest way to get a dataset for training is to download the LLD dataset: https://data.vision.ee.ethz.ch/sagea/lld/

Alternatively, icons can be downloaded seperately and loaded from a folder, the process is described in the iPython notebook.

Pre-Trained Weights

Weights for the results of the improved convolutional GAN and the ResNet WGAN can be downloaded here:

https://drive.google.com/open?id=1FFKdxTGTTUdLZQOeoWkqFb9AegGm3dSF

Commands for the Tensorflow Docker Container

Command for jupyter docker container with current directory as connected drive:

docker run -v $(pwd):/tf/icon-gan -it --rm -p 8888:8888 tensorflow/tensorflow:latest-py3-jupyter

Command with Nvidia GPU enabled, recommended for training:

docker run --runtime=nvidia -v $(pwd):/tf/icon-gan -it --rm -p 8888:8888 tensorflow/tensorflow:latest-gpu-py3-jupyter