Hierarchical Autoregressive Modeling for Neural Video Compression

TL;DR

This paper introduces a hierarchical autoregressive model for neural video compression, connecting autoregressive generative models with video compression tasks, leading to improved rate-distortion performance on large-scale data.

Contribution

It proposes a novel perspective linking autoregressive models to neural video compression, offering new avenues for enhancing compression efficiency.

Findings

Outperforms state-of-the-art neural video compression methods

Achieves better rate-distortion trade-offs on large-scale datasets

Provides a unified stochastic autoregressive framework for video compression

Abstract

Recent work by Marino et al. (2020) showed improved performance in sequential density estimation by combining masked autoregressive flows with hierarchical latent variable models. We draw a connection between such autoregressive generative models and the task of lossy video compression. Specifically, we view recent neural video compression methods (Lu et al., 2019; Yang et al., 2020b; Agustssonet al., 2020) as instances of a generalized stochastic temporal autoregressive transform, and propose avenues for enhancement based on this insight. Comprehensive evaluations on large-scale video data show improved rate-distortion performance over both state-of-the-art neural and conventional video compression methods.