Paper Key : IRJ************599
Author: Aniket Wagh,Shubham Kaware ,Rushikesh Varkale,Abhishek Bhosale
Date Published: 17 Nov 2023
Abstract
This project delves into the intricate realm of generating images through the convergenceof textual descriptions and existing images. Employing the prowess of Generative AdversarialNetworks (GANs), the endeavor addresses the quintessential challenges in AI-fueled imagesynthesis. This venture bears profound significance across the landscape of computer vision,advertising, and entertainment. The primary challenges encompass scarcities in dataset availability,forging meaningful semantic bridges between text and images, and the art of rendering realisminto generated images. Our mission is meticulously honed: fashioning a GAN-centric model thatorchestrates the fusion of text and images, yielding high-caliber, contextually accurate visualmarvels. Beyond catapulting creative workflows and automating image inception, the projectpropels a wave of innovation across a diverse spectrum of industries.The methodology orchestrates the meticulous training of a generator network to conjureimages and a discriminator network to discern authenticity from generated renditions. Guided byiterative training and bolstered by preprocessing techniques, the system acquires the art offabricating images imbued with coherent narratives and aesthetic authenticity. This innovationholds the potential to reframe the contours of image creation, charting a pioneering path withinAI-driven image synthesis. Keywords: GANs (Generative Adversarial Networks), ComputerVision, Generator Network, Discriminator Network, High-Quality Images
DOI LINK : 10.56726/IRJMETS45822 https://www.doi.org/10.56726/IRJMETS45822