A Review of GAN-based Methods for Image Translation and Caption Generation
A. R. Revathi, Swetha Sriram, K M Lohithaa, Sanjana Sudarsan · 2025
The ability of Generative Adversarial Networks (GANs) to produce realistic synthetic data has drawn a lot of attention in recent years. They are a kind of neural network that has two major components: a generator and a discriminator and are trained adversarially to generate data resembling real-world examples. Today, they are commonly used in fields such as medical imaging enhancement, video, speech, and signal processing. In image processing, GANs help address challenges such as face completion, motion image deblurring, and underwater visibility enhancement. This survey paper focuses on advancements in image-to-image and image-to-text transformations, analysing various architectural aspects and their impact on performance based on evaluation parameters. By reviewing qualitative and quantitative results from key research studies, the paper explores existing approaches and provides insights into their applications in GAN-based image translation.