Neural Style Transfer for Picture with Gradient Gram Matrix Description

  • Heng Jin
  • , Tian Wang
  • , Mengyi Zhang
  • , Mingmin Li
  • , Yan Wang
  • , Hichem Snoussi

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Despite the high performance of neural style transfer on stylized pictures, we found that Gatys et al [1] algorithm cannot perfectly reconstruct texture style. Output stylized picture could emerge unsatisfied unexpected textures such like muddiness in local area and insufficient grain expression. Our method bases on original algorithm, adding the Gradient Gram description on style loss, aiming to strengthen texture expression and eliminate muddiness. To some extent our method lengthens the runtime, however, its output stylized pictures get higher performance on texture details, especially in the elimination of muddiness.

Original languageEnglish
Title of host publicationProceedings of the 39th Chinese Control Conference, CCC 2020
EditorsJun Fu, Jian Sun
PublisherIEEE Computer Society
Pages7026-7030
Number of pages5
ISBN (Electronic)9789881563903
DOIs
StatePublished - Jul 2020
Event39th Chinese Control Conference, CCC 2020 - Shenyang, China
Duration: 27 Jul 202029 Jul 2020

Publication series

NameChinese Control Conference, CCC
Volume2020-July
ISSN (Print)1934-1768
ISSN (Electronic)2161-2927

Conference

Conference39th Chinese Control Conference, CCC 2020
Country/TerritoryChina
CityShenyang
Period27/07/2029/07/20

Keywords

  • Gradient Gram matrix
  • art style transfer
  • neural style transfer
  • picture stylize

Cite this