Client-side video colorization using deep neural networks

In this thesis project we look at ways to utilize neural networks and computational frameworks to perform automatic image and video colorization, with the focus on one kind of deep neural network architecture – pix2pix. The project is aimed at modeling colorization process using a deep learning framework called Apache MXNet, training several neural networks on a set of colorful images and using the trained models in a computer browser. The browser can perform image and video colorization in real-time. All the project artifacts are documented and open sourced on the web.