Abstract

This paper proposes the novel Pose Guided Person Generation Network (PG$^2$)that allows to synthesize person images in arbitrary poses, based on an imageof that person and a novel pose. Our generation framework PG$^2$ utilizes thepose information explicitly and consists of two key stages: pose integrationand image refinement. In the first stage the condition image and the targetpose are fed into a U-Net-like network to generate an initial but coarse imageof the person with the target pose. The second stage then refines the initialand blurry result by training a U-Net-like generator in an adversarial way.Extensive experimental results on both 128$\times$64 re-identification imagesand 256$\times$256 fashion photos show that our model generates high-qualityperson images with convincing details.