Text-to-Image Synthesis using GAN


project-image

GAN ... (with source code will update soon...)

Generation of Plausible image from text description as a input prompt. And input image is noise.

       to       


Image dataset used: Oxford 102 flowers

(Python, TensorFlow, DCGAN, RNN, CNN, LSTM, ReLu,)

TASKS/ACHIEVEMENTS:

1. GAN neural networks (Deep Learning)

  • Generator, Discriminator concepts
  • Text embeddings for visual attributes
  • Interpolation of text embeddings
  • Refinements, Loss calculations


Objective:

  • To develop a model to generate plausible images of flower from the detailed text descriptions.


Project details...

Generative Adversarial Neural networks 

- Output as plausible Image of flower when text description of flower is given as input.

Project screenshot

         

Generated Image:

Text: "the petals on this flower are yellow with a red center and pink leafs"

https://lh6.googleusercontent.com/9LBuU4nGdArbKZVU8jvfAdm7K5TPBzHwa14LxNAbMhapTjXzKMRmAAk8r67FQuahoYrAxEphdz2kjEShoDnpx7gsMApVnr7JwILpibJj0RND2SJ-fiDUPjrvEzsdo-fQvNQChD0T

https://lh6.googleusercontent.com/fCHVy_Zn6ZuTLc5ix2nta7Wm1Gh2xVFDCi12cpXrXIFMuqBwprZlbJlpCmBgoTY2w0A1L95mLLKlDV2Cl3hPfKhC-KUlppwB16siPuOUN5zm-hWdwBZGczdDpgzhl_CJ-PdulJd_