• Parmar Digantkumar N

Generative Adversarial Networks (GANs) have a wide range of applications in fields such as computer vision, natural language processing, and speech synthesis. Among these, the most notable achievements have been in image synthesis, particularly in generating deepfake videos. Despite the negative media coverage surrounding deepfakes, they can be valuable in applications like entertainment, customer relations, and assistive care. A significant challenge in creating deepfakes is the need for extensive image training data of the subject, which is not an issue for celebrities with abundant images. However, if only a limited number of training images are available, the deepfake's quality will be poor. Although some media reports claim that a good deepfake can be made with as few as 500 images, in practice, high-quality deepfakes require thousands of images. This requirement contributes to the popularity of deepfakes of celebrities and politicians. In this study, we leverage the ability of GANs to generate images of an individual with varied facial expressions. By using these synthetic, variable-expression training images in smaller quantities, we achieve the production of near-realistic deepfake videos.

Deepfake generation, Generative Adversarial Networks:GANs, Variable face images

" Harnessing GANs for Deepfakes: Minimizing Training Data Requirements", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 2, page no.g770-g774, February-2024, Available :http://www.jetir.org/papers/JETIR2402697.pdf

" Harnessing GANs for Deepfakes: Minimizing Training Data Requirements", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 2, page no. ppg770-g774, February-2024, Available at : http://www.jetir.org/papers/JETIR2402697.pdf