Abstract
Sketching or doodling is a popular creative activity that people engage in.However, most existing work in automatic sketch understanding or generation hasfocused on sketches that are quite mundane. In this work, we introduce twodatasets of creative sketches -- Creative Birds and Creative Creatures --containing 10k sketches each along with part annotations. We propose DoodlerGAN-- a part-based Generative Adversarial Network (GAN) -- to generate unseencompositions of novel part appearances. Quantitative evaluations as well ashuman studies demonstrate that sketches generated by our approach are morecreative and of higher quality than existing approaches. In fact, in CreativeBirds, subjects prefer sketches generated by DoodlerGAN over those drawn byhumans! Our code can be found at https://github.com/facebookresearch/DoodlerGANand a demo can be found at http://doodlergan.cloudcv.org.