During the past few years, video streaming services has developed rapidly around the world. Many popular platforms such as YouTube and Coursera that are often visited are mainly occupied by video contents. Cisco’s study predicts that 82% of entire consumer network traffic will come from video traffic by 2020, which means challenges regarding video storage will arise soon. Therefore, to fulfil storage and streaming demands over the internet, video compression becomes a vital process. On the other hand, neural network has obtained great success in computer vision, image processing and image compression.
In this project, we will i) investigate the methodologies that combine neural networks and video compression methods and ii) try to represent prediction and corresponding residual through the use of neural networks.