Interactive Art Generation with Style Transfer using Image URL

Abstract

This research study presents style transfer between images using VGG19 model, neural style transfer (NST) algorithms for an innovative art generation. The previous art generation methods use VGG16 model which doesn’t predict or create images accurately. As the complexity of images increases, the NST algorithms needs to be trained to improve its performance. To overcome the problems, a convolutional neural network with 19 deep layers is introduced. For the frontend designers and artist, it is a challenging task to create new images for creating web pages. The process involves providing image URLs of two images-content image, style images to transfer styles from one to another. This art generation has emerged as a dynamic and innovative approach for digital art creation which enables individuals to actively participate in artistic process. The architecture adept at preserving content while effectively transferring desired stylistic elements. The aim is to provide a user-friendly interface for the end user which generates many new images in the future. The research study involves tensorflow Hub, which contains pre-processed machine learning algorithms.