Project Abstract

Style Transfer, in the context of Artificial Intelligence (AI), refers to a computational technique that aims to apply the visual style of one image (style image) onto the content of another image (content image). This technology has grown exponentially over the years, being a core principle in photo editing tools, social media filters and marketing.In the rapidly evolving landscape of the modern world, advancements in this technology continue to accelerate and keeping pace has become increasingly challenging for individuals without technical expertise in the field. This project aims to enable the use of the Style Transfer technology to a broader audience, with or without experience in the field, by developing an accessible framework that allows users to Train, Load, Test and Visualise their own work through the use of a Graphical User Interface (GUI). The framework is built around a CycleGAN (Cycle-Consistent Generative Adversarial Network), which is a specialised GAN model for Image-to-Image translation tasks. With this model, I plan to capture the style of iconic painters from the past, like Monet or Van Gogh, and apply their artistic style to contemporary pictures. I will train different models and make them available for the users to play with the technology.By developing this tool, I intend to create more than an accessible framework and contribute to the convergence of art and technology.

Keywords: Artificial Intelligence, Style Transfer, Generative Adversarial Network

 Conference Details

Session: Poster Session A at Poster Stand 36

Location: Sir Stanley Clarke Auditorium at Tuesday 7th 13:30 – 17:00

Markers: Amjad Amjad, Gary Tam

Course: BSc Computer Science FI, 3rd Year

Future Plans: I have a job lined-up