Friday 11:30 a.m.–noon in Grand Ballroom B

DragonPaint – Bootstrapping Small Data to Color Cartoons

Gretchen Greene


The creation of sufficient quantities of labeled training data is one of the biggest challenges for machine learning applications, especially when the data itself must be created, not just labeled. DragonPaint presents a generalizable strategy for minimizing the manual creation of data using rule based algorithms to automate the creation of a restricted subset of data and then bootstrapping to the automated creation of unrestricted (rule breaking) training and test data. A gentle introduction to computer vision, graphics and machine learning, we use Python and geometry to build an image data set for a TensorFlow model.