Python-Interface-to-Create-Handwritten-dataset

Python Interface to Create Handwritten dataset

Original article and full tutorial : Build Your Fully Configurable Interface to Create MNIST-like Dataset with Python in 10 Minutes

Requirements

pip3 install -r requirements.txt
or 
pip3 install numpy opencv-python

Run the programm

python3 main.py -c path/to/the/config/file.json
or 
python3 main.py -config path/to/the/config/file.json

Configuration JSON file

• input → H (int): height of the drawing image
• input → W (int): width of the drawing image
• input → thickness (float): proportionnal thickness according to the diagonal of the image size
• output → H (int): height of the output image
• output → W (int): width of the output image
• process → volume (int): number of sample by class we can to get at the end
• process → selection (str): process of selection. can be RANDOM|CLASSBYCLASS|ROTATE
• process → display_output (bool): display the ouput in a window or not
• storage → root (str): path to the root folder
• storage → by_class_name (bool): store classes in subfolders
• line → thickness (int or list of int): proportionnal thickness according to the diagonal of the input size
• line → range_value (int or list of int): random range for line value when drawing if list else value for the line
• line → fading (float): value of the border fade for a linear fading
• interpolation → method (str or list of str): OpenCV interpolation for resizing to ouput. Can be INTER_NEAREST|INTER_LINEAR|INTER_AREA|INTER_CUBIC|INTER_LANCZOS4|RANDOM
• classes (list of str): contains the name of all the caracters

Enjoy !