SIGGRAPH Asia 2021

Pose with Style: Detail-Preserving Pose-Guided Image Synthesis with Conditional StyleGAN


Badour AlBahar1,2Jingwan Lu3Jimei Yang3Zhixin Shu3Eli Shechtman3Jia-Bin Huang1,4
1Virginia Tech2Kuwait University3Adobe Research4University of Maryland College Park




Results on DeepFashion StylePoseGAN train/test split



Reference

  • GFLA: Deep Image Spatial Transformation for Person Image Generation.
  • ADGAN: Controllable Person Image Synthesis with Attribute-Decomposed GAN.
  • PATN: Progressive Pose Attention for Person Image Generation.
  • StylePoseGAN: Style and Pose Control for Image Synthesis of Humans from a Single Monocular View.