Abstract
Majority of researches are taking place in the image-processing dimension as the demand for image processing applications is widely increasing. Generative adversarial network is a state-of-the-art image-processing technology in recent improvement and development in the field of image processing. Generative adversarial network is a neural network architecture which is used as a generative model. Generative adversarial network is popularly known as GANs. GANs can be designed for generating scores of images from the data sets, text to image conversion, face aging, animations, storytelling and so on. GANs can also be used in the field of signal communication to attain optimization in combinational design. Based upon their design and applications, a number of variants of GANs have come into existence. The various classes of GANs include semantic enhancement GANs, resolution enhancement GANs, diversity enhancement GANs and motion enhancement GANs. Popular examples of GANS are deep convolution generative adversarial network (DCGAN), conditional generative adversarial network (cGAN), stack GAN, (discover cross-domain relations with generative adversarial networks (Disco GAN), InfoGAN Age-cGAN and so on. Basically, any GAN framework works on the adversarial principle of evaluating generative models. Models named generator and discriminator will be trained simultaneously. The role of generator is to synthesize a series of outputs and the role of discriminator is to differentiate the currently generated output from the previous outputs. This generation and discrimination process iterates until the discriminator declares the current output to be 100% different from the previous outputs. This chapter thoroughly introduces GANs and their latest trends for performing image processing and computer vision.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call