Deep Drawing

Unleash your creativity using AI
Try yourself »

Table of Contents

1. About The Project
   • Built With
2. Getting Started
   • Prerequisites
   • Installation
3. Examples
4. Contact
5. Acknowledgments

About The Project

Can an algorithm learn to draw? Since the DeepMind deep learning network won against the Go player Lee Sedol in 2016, it seems that any highly complex task can be invested by these computing techniques. Now, traditionally, drawing is a cognitive task that is complex because it combines the mind with the mastery of a gesture. From very large sets of drawings, GANs, new deep learning architectures, can generate realistic or convincing graphics. Other questions then arise: Can we create a program that is creative (or aids creativity)? And can we collaborate with it in order to invent new methods of creation? These are the questions that our project tries to answer.

(back to top)

Built With

• Tensorflow
• TensorflowJs
• Python

(back to top)

Getting Started

There are a few things to do first if you want to try the 'non web' version of our project

Prerequisites

You will have to install some software in order to run our project:

• Python
• Anaconda
• VSCode (Optional, you can use Anaconda if you prefer)

Installation

Once everything above is set up you will have to run several commands in your terminal:

1. First we create a virtual environment:

  conda create --name NAMEOFYOURENVIRONMENT python=3.6

2. We then activate this new environment:

   conda activate NAMEOFYOURENVIRONMENT

3. Now you can either install pip dependencies by hand or you can write the following command in your terminal:

   pip install -r requirements.txt

4. If you prefer going by hand enter all this commands in your terminal:

pip install tensorflow-directml

pip install tensorflow_hub

pip install ipykernel

pip install matplotlib

pip install notebook

pip install pandas

pip install imageio

pip install graphviz

pip install scipy

pip install pydot

pip install opencv-python

5. You can now run the notebook in your preferred IDE

Run with Google Colab:

If you prefer using Google Colab to train the model:

1. Open the following link and run the notebook, everything is already set-up: Google Colab Notebook

2. Or create a notebook yourself and copy these commands:

!git clone 'https://github.com/leogenot/GenerativeDeepDrawing/'

%cd GenerativeDeepDrawing

3. Then copy and paste each cells of the wgan_gp.ipynb notebook by hand.

(back to top)

Examples

(back to top)

Contact

Léo Genot - LinkedIn
Valentin Guiller - LinkedIn
Adam Ferreira - LinkedIn

Vincent Nozick - LinkedIn
Gaëtan Robillard - LinkedIn

IMAC engineering school - Website

Project Link: Github
Website Link: Deep Drawing

(back to top)

Acknowledgments

Some ressources that were helpful for the project:

• David Foster Book
• Keras Documentation
• Tensorflow Documentation

(back to top)