This codes modify the implementation of the ICCV-2017 paper "GANs for Biological Image Synthesis" for generating images of Yeast cell.
This code contains the following code related to generating fakes images of yeast cell:
- implementation of WGAN-GP
The biogan code is released under Apache v2 License allowing to use the code in any way you want.
###Original images of yeast cell:
###Generated fake images of yeast cell after 69K iterations:
If you are using this software please cite the following paper in any resulting publication:
Anton Osokin, Anatole Chessel, Rafael E. Carazo Salas and Federico Vaggi, GANs for Biological Image Synthesis, in proceedings of the International Conference on Computer Vision (ICCV), 2017.
@InProceedings{osokin2017biogans,
author = {Anton Osokin and Anatole Chessel and Rafael E. Carazo Salas and Federico Vaggi},
title = {{GANs} for Biological Image Synthesis},
booktitle = {Proceedings of the International Conference on Computer Vision (ICCV)},
year = {2017} }
This software was written for python v3.6.1, pytorch v0.2.0 (earlier version won't work; later versions might face some backward compatibility issues, but should work), torchvision v0.1.8 (comes with pytorch). Many other python packages are required, but the standard Anaconda installation should be sufficient.
We tested this code on Ubuntu 16.04 and Macbook, but in case of Macbook we encountered issues related to symbolic linking. So it's better to test and run this code on Linux environment only. Also GPU is preferred so that code can be run for enough iterations to get promissing images.
This code is modified version of Biogan research paper by Anton Osokin, Anatole Chessel, Rafael E. Carazo Salas and Federico Vaggi, GANs for Biological Image Synthesis, in proceedings of the International Conference on Computer Vision (ICCV), 2017.
The modified code contains training and testing data to generate fake single channel images of yeast cell.
Note that rerunning all the experiements would require significant computational resources. We recommend using a cluster of GPU if you want to do that.
Get the code
git clone https://github.com/geekyspartan/biogans.git
Mark the root folder for the code
cd biogans
export ROOT_BIOGANS=`pwd`
cd $ROOT_BIOGANS/experiments/models_6class_joint
./train_size-48-80_6class_wgangp-adam.sh