README.md

scarabocchi

Barebone tools for interactive image scribbles on python

Requirements

• numpy
• matplotlib
• skimage

Usage

Run the annotation tools:

from scarabocchi.scarabocchi import scribbles_tools2d
from skimage.data import astronaut

# Simple annotator example
annotator = scribbles_tools2d(img=astronaut(), figsize=(10, 10))

• Paint scribbles by mouse click and move on the image.
• Use digits or up-down arrows to select new labels.
• remove scribble with r.
• new random color map with t.

Annotator

The annotator object contains all labels information:

• a label mask with same spatial dimension as the initial image.

annotator.mask

• a dictionary with all annotation information and pixel-wise segmentation for each object.

# Example for segment "1"
annotation = annotator.labels["1"]

# Scribbles
annotation.xdata, annotation.ydata

# Object segmentation
annotation.xsegdata, annotation.ysegdata

Segmentation

Segmentation is performed with a simple seeded watershed, for custom segmentation algorithm just override the following method:

annotator.update_segmentation() 

Jupyter notebook

To get scarabocchi works on jupyter notebooks to use the following magic commands

%matplotlib notebook