Auto_painter_demo

The code for building a web demo of Auto_painter

Introduction

This is the web demo code of AutoPainter. See the demo: https://youtu.be/g9rf-YFGgbg

There are three models, including a demo model trained on Minions, a model without color control, and a model with color control.

The sketch can be uploaded or generated by an uploaded RGB image, which means you can generate a cartoon portrait of yourself~

Enjoy this web demo. You can also build other GAN demo with the project.

Requirements

python3.5

tensorflow1.4

tornado

werkzeug

flask

urllib

skimage

PIL

You can 'pip install' all the dependency based on Anaconda

Quick start

1. Download three pre-train models from this link: https://cloudstor.aarnet.edu.au/plus/s/ZEhW8CVc73IHC15
2. unzip it, and place it under the folder 'Auto_painter_demo', like

• 'Auto_painter_demo/model1'
• 'Auto_painter_demo/model2'
• 'Auto_painter_demo/model3'

3. RUN python app2.py

It will open a web demo at your localhost:5001. You can change this ip by modifying the code.

Reference

If you find this code useful, you can cite the paper:

@article{liu2018auto, title={Auto-painter: Cartoon image generation from sketch by using conditional Wasserstein generative adversarial networks}, author={Liu, Yifan and Qin, Zengchang and Wan, Tao and Luo, Zhenbo}, journal={Neurocomputing}, volume={311}, pages={78--87}, year={2018}, publisher={Elsevier} }