Rethinking Learned Image Compression: Context is All You Need

TL;DR

This paper investigates the limits of learned image compression (LIC), analyzing the role of scaling parameters and overfitting, and demonstrates state-of-the-art PSNR performance with significant BD-RATE gains over VVC.

Contribution

It reveals that overfitting can serve as an effective context in LIC and proposes optimized scaling of components to push compression performance boundaries.

Findings

Overfitting can enhance context modeling in LIC.

Scaling encoder, decoder, and context model improves PSNR.

Achieves 14.39% BD-RATE gain over VVC.

Abstract

Since LIC has made rapid progress recently compared to traditional methods, this paper attempts to discuss the question about 'Where is the boundary of Learned Image Compression(LIC)?'. Thus this paper splits the above problem into two sub-problems:1)Where is the boundary of rate-distortion performance of PSNR? 2)How to further improve the compression gain and achieve the boundary? Therefore this paper analyzes the effectiveness of scaling parameters for encoder, decoder and context model, which are the three components of LIC. Then we conclude that scaling for LIC is to scale for context model and decoder within LIC. Extensive experiments demonstrate that overfitting can actually serve as an effective context. By optimizing the context, this paper further improves PSNR and achieves state-of-the-art performance, showing a performance gain of 14.39% with BD-RATE over VVC.