Deep Generative Video Compression

TL;DR

This paper introduces a deep generative VAE-based approach for video compression, demonstrating competitive results on public datasets and exceptional performance on specialized content.

Contribution

It presents a novel end-to-end deep generative model for video compression that combines variational autoencoders with neural image compression techniques.

Findings

Competitive rate-distortion performance on public videos

Exceptional compression on specialized content

Effective end-to-end training for temporal sequences

Abstract

The usage of deep generative models for image compression has led to impressive performance gains over classical codecs while neural video compression is still in its infancy. Here, we propose an end-to-end, deep generative modeling approach to compress temporal sequences with a focus on video. Our approach builds upon variational autoencoder (VAE) models for sequential data and combines them with recent work on neural image compression. The approach jointly learns to transform the original sequence into a lower-dimensional representation as well as to discretize and entropy code this representation according to predictions of the sequential VAE. Rate-distortion evaluations on small videos from public data sets with varying complexity and diversity show that our model yields competitive results when trained on generic video content. Extreme compression performance is achieved when training the model on specialized content.