/Text-to-Image-Synthesis

PyTorch Implementation of the paper - 'Generative Adversarial Text to Image Synthesis' from ICML 2016 https://arxiv.org/abs/1605.05396

Primary LanguagePythonMIT LicenseMIT

Text-to-Image-Synthesis

PyTorch Implementation of the paper - 'Generative Adversarial Text to Image Synthesis' from ICML 2016 [ https://arxiv.org/abs/1605.05396 ]

Please refer the report.pdf for overview of the process and details regarding the implementation.

The original paper implemented 4 methods

  • GAN
  • GAN-CLS [Matching-aware discriminator]
  • GAN-INT [Learning with manifold interpolation]
  • GAN-INT-CLS [Combination of above two methods]

However, this code only contains the Pytorch implementation of GAN-CLS.

Please refer the original paper [ https://arxiv.org/abs/1605.05396 ] for the details of other methods.