Multimodal Text-2-Face Network for Human Portrait Generation

May 15, 2019
Presentation
Face generation based on text prompt

  • Built an automated system to generate human faces based on textual descriptions, using Generative Adversarial Networks.
  • Implemented a combination of StackGAN + ProgressiveGAN architectures to generate 64 x 64 images.
  • Utilized the infersent module in combination with the GloVe model for text embedding.

ML