pktiuk/evolutionary-painter

Simple genetic Python C++ app recreating image by drawing multiple rectangles

Rectangles Fitting image

Application recreating image by drawing multiple rectangles. To achieve it we are using genetic algorithm.

It provides simple Python library with binds to C++ SFML.

Original

Output

Video Example

Preparing

Install dependencies

Packages

Required to build main executable.

sudo apt-get install cmake libboost-python-dev libsfml-dev  python3 python3-pip jupyter-notebook

Python packages

(python part is optional, but it is recommended for analyzing algorithm)

pip3 install numpy matplotlib notebook

Download and build

git clone https://gitlab.com/pktiuk/evolutionary-painter.git
cd evolutionary-painter
mkdir build
cd build
cmake ..
cmake --build .

Using app

There are two ways of running this app:

• C++-based - it is used to generate output image using simple C++ executable.

Usage:    evolving-rectangles [OPTIONS]
          -i  input file (if not described program will show selection window)
          -p  population size (default 30)
          -g  gene pool, maximum number of rectangles on picture (default 1000)
          -c  end simulation after given amount of generations (empty or 0, no restriction)
          -t  run in terminal mode only (without drawing window)
          -h  print help

• Python-based - it is recommended for more research and test-focused usage. It uses C++ to python bindings.
  It can be launched imported from file BindWrapper.py (you can just run this file and after importing everything you will be left in the interactive mode). You can also use Jupyter Notebook for easy tests. Example notebook - it can be launched via jupyter notebook ./notebook.ipynb