Image Generation Based on Text Using BERT And GAN Model
Mallaiahgari Rohith, L. Pallavi, Kogila Shirisha, Munukoti Sanjay, V. Sathya Priya · 2023
One of the most challenging and important problems in deep learning is creating visuals using a text description. The sub-domain of text-to-image generation is text- to-face image generation. The end objective is to deliver the image utilizing the client-determined face portrayal. Our proposed paradigm includes both images and text. There are two phases to the proposed work. The conversion of the text into semantic features is demonstrated in the first phase. These semantic features have been used in the second phase to train the image decoder to produce accurate natural images. Creating an image based on a written description is more applicable to public safety responsibilities. The fully trained GAN that has been proposed outperformed by producing high-quality images from the input phrase.