Good, Cheap, and Fast: Overfitted Image Compression with Wasserstein Distortion

TL;DR

This paper introduces an overfitted image compression method optimized for Wasserstein Distortion, achieving high perceptual quality with minimal computational cost and outperforming existing perceptual metrics in predicting human ratings.

Contribution

It demonstrates that modeling visual perception with Wasserstein Distortion enables efficient compression with high perceptual quality and low complexity, surpassing traditional perceptual metrics.

Findings

Wasserstein Distortion outperforms LPIPS in human rating correlation.

The proposed codec requires less than 1% of MACs compared to generative models.

WD achieves over 94% Pearson correlation with human Elo scores.

Abstract

Inspired by the success of generative image models, recent work on learned image compression increasingly focuses on better probabilistic models of the natural image distribution, leading to excellent image quality. This, however, comes at the expense of a computational complexity that is several orders of magnitude higher than today's commercial codecs, and thus prohibitive for most practical applications. With this paper, we demonstrate that by focusing on modeling visual perception rather than the data distribution, we can achieve a very good trade-off between visual quality and bit rate similar to "generative" compression models such as HiFiC, while requiring less than 1% of the multiply-accumulate operations (MACs) for decompression. We do this by optimizing C3, an overfitted image codec, for Wasserstein Distortion (WD), and evaluating the image reconstructions with a human rater study, showing that WD clearly outperforms LPIPS as an optimization objective. The study also reveals that WD outperforms other perceptual metrics such as LPIPS, DISTS, and MS-SSIM as a predictor of human ratings, remarkably achieving over 94% Pearson correlation with Elo scores.