Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis

Chuan Li, Michael Wand
1/18/2016
cs.CV

Official code: chuanli11/CNNMRF (859 stars, MIT, last commit 2019-10-12).

Abstract

This paper studies a combination of generative Markov random field (MRF) models and discriminatively trained deep convolutional neural networks (dCNNs) for synthesizing 2D images. The generative MRF acts on higher-levels of a dCNN feature pyramid, controling the image layout at an abstract level. We apply the method to both photographic and non-photo-realistic (artwork) synthesis tasks. The MRF regularizer prevents over-excitation artifacts and reduces implausible feature mixtures common to previous dCNN inversion approaches, permitting synthezing photographic content with increased visual plausibility. Unlike standard MRF-based texture synthesis, the combined system can both match and adapt local features with considerable variability, yielding results far out of reach of classic generative MRF methods.

View on arXivView PDF

Code Implementations(1)

chuanli11/CNNMRFOfficial100%
MIT

code for paper "Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis"

859144Lua, ShellNov 22, 20157 years agoMIT

Similar Papers

Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks

86% match1 repo2912016

Learning Layout and Style Reconfigurable GANs for Controllable Image Synthesis

84% match1 repo532020

Image Synthesis under Limited Data: A Survey and Taxonomy

84% match1 repo872023

SinGAN: Learning a Generative Model from a Single Natural Image

84% match1 repo3.3k2019

Neural Photo Editing with Introspective Adversarial Networks

84% match1 repo2.1k2016

Scaling Non-Parametric Sampling with Representation

83% match2025

Cite this paper

@article{li2016combining,
  title  = {Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis},
  author = {Chuan Li and Michael Wand},
  year   = {2016},
  eprint = {1601.04589},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1601.04589}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1601.04589.svg)](https://codeofpaper.com/paper/1601.04589)

Discussion