josephrocca / tfjs-gans

GAN models in the browser with tensorflow.js

GAN models in the browser with tensorflow.js

• Demo: https://josephrocca.github.io/tfjs-gans/index.html
• See also: https://github.com/josephrocca/tfjs-acgan-browser-training

Setup

Download this repo and serve the contents of the folder with a static webserver of your choice. For example:

git clone https://github.com/josephrocca/tfjs-gans
cd tfjs-gans
# install deno if you don't have it: https://deno.land/
deno run --allow-net --allow-read=. https://raw.githubusercontent.com/josephrocca/denoSimpleStatic/master/main.ts

The server will say something like "Start listening on 127.0.0.1:8000", and so you'd open up 127.0.0.1:8000 in your browser.

Training a model

You can use this colab notebook to train your own DCGAN, or you can convert a ProGAN/PGGAN, or BigGAN model (among others) from tfhub.dev to tfjs format by opening a new Google Colab notebook and running the following code:

!pip install tensorflowjs[wizard].
!tensorflowjs_wizard

Enter a tfhub URL like https://tfhub.dev/google/progan-128/1 when prompted, and use all the default options. Then download the model files. More detailed instructions are in the tfjs-converter repo.