Paper Title
Keras Implementation of Neural Style Transfer

Abstract
This paper uses machine learning programs to transfer everyday images to art style images by using a style image to get the transferred style and a content image we want to transfer the style onto. This project can show how machine learning techniques do well in the use of images and how artificial intelligence can achieve great success in art and creativity. Current style transfer techniques such as Prisma emphasized more on the color and layouts of style image. Hence, the synchronized image might lose color, and the content image's outline gets distorted. This project explores a process that can preserve content images' elements and apply artistic styles from style images. The neural style algorithm is very successful in discovering the style of famous pieces of art. There have been attempts to cover this style in images on various levels of success. Keywords - Neural Style Transfer, CNN, Deep Neural Network, Keras, Content Loss, Style Loss, Feature Maps, Gram Matrix, Non-photorealistic Rendering, Convolutional Layers, Deep Learning.

Author - Abhilasha Sharma, Ritik Handa, Rakshit, Raghav Punjani

Published : Volume-8,Issue-6  ( Jun, 2021 )

	|		PDF	|	Viewed - 33	|	Published on 2021-10-18