Exploring the structure of a real-time, arbitrary neural artistic stylization network

   Abstract