Texture Synthesis Using Convolutional Neural Networks
Leon A. Gatys Affiliation: Centre for Integrative Neuroscience, University of Tübingen, Germany Affiliation: Bernstein Center for Computational Neuroscience, Tübingen, Germany Affiliation: Graduate School of Neural Information Processing, University of Tübingen, Germany Email: Alexander S. Ecker Affiliation: Centre for Integrative Neuroscience, University of Tübingen, Germany Affiliation: Bernstein Center for Computational Neuroscience, Tübingen, Germany Affiliation: Max Planck Institute for Biological Cybernetics, Tübingen, Germany Affiliation: Baylor College of Medicine, Houston, TX, USA Matthias Bethge Affiliation: Centre for Integrative Neuroscience, University of Tübingen, Germany Affiliation: Bernstein Center for Computational Neuroscience, Tübingen, Germany Affiliation: Max Planck Institute for Biological Cybernetics, Tübingen, Germany
Abstract
Here we introduce a new model of natural textures based on the feature spaces of convolutional neural networks optimised for object recognition. Samples from the model are of high perceptual quality demonstrating the generative power of neural networks trained in a purely discriminative fashion. Within the model, textures are represented by the correlations between feature maps in several layers of the network. We show that across layers the texture representations increasingly capture the statistical properties of natural images while making object information more and more explicit. The model provides a new tool to generate stimuli for neuroscience and might offer insights into the deep representations learned by convolutional neural networks.
原文 arXiv:1505.07376;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1505.07376v3