eduzhai > Applied Sciences > Engineering >

XingGAN for Person Image Generation

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 22 pages

Abstract: We propose a novel Generative Adversarial Network (XingGAN or CrossingGAN)for person image generation tasks, i.e., translating the pose of a given personto a desired one. The proposed Xing generator consists of two generationbranches that model the person s appearance and shape information,respectively. Moreover, we propose two novel blocks to effectively transfer andupdate the person s shape and appearance embeddings in a crossing way tomutually improve each other, which has not been considered by any otherexisting GAN-based image generation work. Extensive experiments on twochallenging datasets, i.e., Market-1501 and DeepFashion, demonstrate that theproposed XingGAN advances the state-of-the-art performance both in terms ofobjective quantitative scores and subjective visual realness. The source codeand trained models are available at this https URL.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...