My name is Shayne D'Lima and I am a final year student at Monash University studying a B. Eng (Hons.)/ B. Commerce. For my Final Year Project, I have been working alongside Dr. Mehrtash Harandi to develop a web application that allows users to get a closer look into state-of-the-art research in the field of computer vision and machine learning when it comes to enhancing images.
This web application focusses on the concept of Super Resolution. Super Resolution involves the process of training a Neural Network to increase the resolution of images by inputting a lower resolution image into the network and training the network to output a higher resolution version of that image.I'm keen to try it out!
The main source of inspiration for this project came from "Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network" (Ledig et al., 2017). This paper investigates the use of Generative Adversarial Networks (GANs) to increase the resolution of images, with impressive results. A GAN is a machine learning framework in which two opposing networks, the generator and the discriminator, are trained simultaneously. The generator is trained to generate data that is similar to the data that is fed into it. On the other hand, the discriminator is tasked with determining whether the data it receives is from the actual training set or if it is from the generator.
An implementation of this research paper was made publicly available on Github by sgrvinod. This served as an amazing starting point for this application; adapting and building on this code allowed me to train this network on various datasets and explore other ML concepts such as image inpainting.