Instance Normalization: The Missing Ingredient for Fast Stylization

Code

• github.com/DmitryUlyanov/texture_nets
  OfficialIn papertorch★ 0
• github.com/labmlai/annotated_deep_learning_paper_implementations
  pytorch★ 66,103
• github.com/hollygrimm/cyclegan-keras-art-attrs
  none★ 5
• github.com/tbullmann/imagetranslation-tensorflow
  tf★ 0
• github.com/cryu854/FastStyle
  tf★ 0
• github.com/MarvinLavechin/imagetranslation-tensorflow
  tf★ 0
• github.com/riven314/PerceptualLoss-FastAI
  pytorch★ 0
• github.com/milmor/perceptual-losses-neural-style
  tf★ 0
• github.com/the-super-toys/glimpse-models
  pytorch★ 0
• github.com/brightyoun/Video-Style-Transfer
  pytorch★ 0

Abstract

It this paper we revisit the fast stylization method introduced in Ulyanov et. al. (2016). We show how a small change in the stylization architecture results in a significant qualitative improvement in the generated images. The change is limited to swapping batch normalization with instance normalization, and to apply the latter both at training and testing times. The resulting method can be used to train high-performance architectures for real-time image generation. The code will is made available on github at https://github.com/DmitryUlyanov/texture_nets. Full paper can be found at arXiv:1701.02096.

Tasks

Reproductions