Learned Video Compression for YUV 4:2:0 Content Using Flow-based Conditional Inter-frame Coding

TL;DR

This paper introduces a flow-based conditional inter-frame coding framework for YUV 4:2:0 video compression, achieving better PSNR-YUV than x265 on certain datasets, with room for further improvements.

Contribution

It presents a novel flow-based inter-frame coder and a variable-rate coding strategy tailored for YUV 4:2:0 content, addressing limitations of residual coding and RGB optimization in prior models.

Findings

Outperforms x265 on UVG and MCL-JCV datasets in PSNR-YUV.

Shows potential for improvement on challenging datasets with increased model capacity.

Highlights the importance of inter-frame coding at large GOP sizes.

Abstract

This paper proposes a learning-based video compression framework for variable-rate coding on YUV 4:2:0 content. Most existing learning-based video compression models adopt the traditional hybrid-based coding architecture, which involves temporal prediction followed by residual coding. However, recent studies have shown that residual coding is sub-optimal from the information-theoretic perspective. In addition, most existing models are optimized with respect to RGB content. Furthermore, they require separate models for variable-rate coding. To address these issues, this work presents an attempt to incorporate the conditional inter-frame coding for YUV 4:2:0 content. We introduce a conditional flow-based inter-frame coder to improve the inter-frame coding efficiency. To adapt our codec to YUV 4:2:0 content, we adopt a simple strategy of using space-to-depth and depth-to-space conversions. Lastly, we employ a rate-adaption net to achieve variable-rate coding without training multiple models. Experimental results show that our model performs better than x265 on UVG and MCL-JCV datasets in terms of PSNR-YUV. However, on the more challenging datasets from ISCAS'22 GC, there is still ample room for improvement. This insufficient performance is due to the lack of inter-frame coding capability at a large GOP size and can be mitigated by increasing the model capacity and applying an error propagation-aware training strategy.