作者: Matthias Bethge , Leon A. Gatys , Alexander S. Ecker
DOI:
关键词: Class (computer programming) 、 Perception 、 Quality (philosophy) 、 Computer science 、 Path (graph theory) 、 Artificial neural network 、 Algorithm 、 Facial recognition system 、 Visual perception 、 Object (computer science)
摘要: In fine art, especially painting, humans have mastered the skill to create unique visual experiences through composing a complex interplay between content and style of an image. Thus far algorithmic basis this process is unknown there exists no artificial system with similar capabilities. However, in other key areas perception such as object face recognition near-human performance was recently demonstrated by class biologically inspired vision models called Deep Neural Networks. Here we introduce based on Network that creates artistic images high perceptual quality. The uses neural representations separate recombine arbitrary images, providing algorithm for creation images. Moreover, light striking similarities performance-optimised networks biological vision, our work offers path forward understanding how perceive imagery.