Image processing using deep learning
Loading...
Files
Date
2020
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
UMT.Lahore
Abstract
Entertainment industry is the most emerging and promising field of the time in which
competition is increasing with every day. Kid’s content of entertainment covers a sufficient
part of industry’s revenue. Different kinds of cartoons and animated shapes are the basic need
of kid’s content to capture their attention for every new cartoon series. A professional artist is
required in order to make new cartoons which is time consuming and quite expensive. To tackle
the aforementioned problem we are presenting an artificial intelligence based solution to
generate different types of cartoon without involvement of man force, which is challenging and
valuable in computer graphics and computer vision. Recent research showed that GAN
(Generative Adversarial Network) produces the cartoon with highly visual fidelity and have
achieved great success. GAN based methodology provide automatic cartoon generation. The
deep generative networks have exhibited a remarkable capability in cartoon generation. An
approach for GAN’s Latent vector optimization to generate more realistic cartoons is
introduced in this research. The dataset used for this research has been taken from kaggle to
train our model of GAN as it requires a lot of data to learn. Moreover, an adversarial learning
technique is presented to simultaneously to train a generator and parallel discriminators, The
evaluation of objective of proposed method demonstrate the following results proposed
framework produced more realistic results by comparing the state-of-art and ground truth.