GENERATION OF ANIME FACES USING GANs

Abstract

Generative Adversarial Networks are a topic of interest in the world of DNN (Deep Neural Networks). GAN frameworks were designed by Ian Goodfellow and his colleagues in June 2014. This idea took the world of deep learning by storm. In GAN two neural networks fight against each other, giving feedback to each other in the process, and eventually become very good at a particular task. This idea was so powerful that lots of applications of GANs emerged. Due to GANs, many futuristic concepts came to life. One such cool application is generating real-looking fake images. These images did not exist before but were completely formed by the Generative Adversarial Network. Many more applications like Text-to-image translation, video prediction, face aging, etc. were made possible using GANs. In this report, I will specifically present anime faces generation using GANs.