Learning Frequency-Specific Quantization Scaling in VVC for Standard-Compliant Task-driven Image Coding

TL;DR

This paper introduces a data-driven method to optimize frequency-specific quantization scaling lists in VVC for neural network-based image analysis, achieving significant bitrate savings over standard VVC and human-visual-system-based scaling.

Contribution

It presents a novel approach to derive optimal scaling lists for neural networks in VVC, improving compression efficiency for task-driven image coding.

Findings

Achieves up to 8.9% bitrate savings over standard VVC.

Outperforms scaling lists optimized for human visual perception.

Provides publicly available scaling lists for practical use.

Abstract

Today, visual data is often analyzed by a neural network without any human being involved, which demands for specialized codecs. For standard-compliant codec adaptations towards certain information sinks, HEVC or VVC provide the possibility of frequency-specific quantization with scaling lists. This is a well-known method for the human visual system, where scaling lists are derived from psycho-visual models. In this work, we employ scaling lists when performing VVC intra coding for neural networks as information sink. To this end, we propose a novel data-driven method to obtain optimal scaling lists for arbitrary neural networks. Experiments with Mask R-CNN as information sink reveal that coding the Cityscapes dataset with the proposed scaling lists result in peak bitrate savings of 8.9 % over VVC with constant quantization. By that, our approach also outperforms scaling lists optimized for the human visual system. The generated scaling lists can be found under https://github.com/FAU-LMS/VCM_scaling_lists.