BVI-AOM: A New Training Dataset for Deep Video Compression Optimization

TL;DR

The paper introduces BVI-AOM, a comprehensive and flexible training dataset with diverse video content, designed to improve deep video compression models, leading to measurable bitrate savings over existing datasets.

Contribution

It presents a new large-scale, diverse, and openly licensed training dataset for deep video coding, enhancing model generalization and performance.

Findings

BVI-AOM achieves up to 0.29% PSNR-Y bitrate savings.

BVI-AOM achieves up to 2.98% VMAF bitrate savings.

Dataset covers a wide range of resolutions and content types.

Abstract

Deep learning is now playing an important role in enhancing the performance of conventional hybrid video codecs. These learning-based methods typically require diverse and representative training material for optimization in order to achieve model generalization and optimal coding performance. However, existing datasets either offer limited content variability or come with restricted licensing terms constraining their use to research purposes only. To address these issues, we propose a new training dataset, named BVI-AOM, which contains 956 uncompressed sequences at various resolutions from 270p to 2160p, covering a wide range of content and texture types. The dataset comes with more flexible licensing terms and offers competitive performance when used as a training set for optimizing deep video coding tools. The experimental results demonstrate that when used as a training set to optimize two popular network architectures for two different coding tools, the proposed dataset leads to additional bitrate savings of up to 0.29 and 2.98 percentage points in terms of PSNR-Y and VMAF, respectively, compared to an existing training dataset, BVI-DVC, which has been widely used for deep video coding. The BVI-AOM dataset is available at https://github.com/fan-aaron-zhang/bvi-aom