/SPADE

"Semantic Image Synthesis with Spatially-Adaptive Normalization" paper implementation

Primary LanguageJupyter Notebook

SPADE

"Semantic Image Synthesis with Spatially-Adaptive Normalization" unofficial paper implementation using fastai library. Please find the associated blog here.

Requirements

  • fastai=1.0.39
  • pytorch=1.0.0

TODO

  • Multimodal image synthesis
  • Perceptual loss
  • Embedding Space Semantic Maps
  • Spade Multiscale Discriminator
  • pix2pixHD loss function (hinge version)
  • feature matching loss
  • Spade Generator
  • GANLearner according to SPADE

Citation

  • Taesung Park, Ming-Yu Liu, Ting-Chun Wang: “Semantic Image Synthesis with Spatially-Adaptive Normalization”, 2019, CVPR 2019; arXiv:1903.07291.