Observer Dependent Lossy Image Compression

TL;DR

This paper introduces a unified approach to image compression that considers both human visual perception and machine classification, enabling optimized compression tailored to different observers.

Contribution

It proposes a family of loss functions that interpolate between perceptual quality and classification accuracy, advancing the integration of human and machine-centric image compression.

Findings

Perceptual loss functions preserve classification accuracy better than traditional codecs.

Compressing ImageNet to 0.25 bpp reduces classification accuracy by only 2%.

Trade-off exists between human visual quality and classification performance.

Abstract

Deep neural networks have recently advanced the state-of-the-art in image compression and surpassed many traditional compression algorithms. The training of such networks involves carefully trading off entropy of the latent representation against reconstruction quality. The term quality crucially depends on the observer of the images which, in the vast majority of literature, is assumed to be human. In this paper, we aim to go beyond this notion of compression quality and look at human visual perception and image classification simultaneously. To that end, we use a family of loss functions that allows to optimize deep image compression depending on the observer and to interpolate between human perceived visual quality and classification accuracy, enabling a more unified view on image compression. Our extensive experiments show that using perceptual loss functions to train a compression system preserves classification accuracy much better than traditional codecs such as BPG without requiring retraining of classifiers on compressed images. For example, compressing ImageNet to 0.25 bpp reduces Inception-ResNet classification accuracy by only 2%. At the same time, when using a human friendly loss function, the same compression system achieves competitive performance in terms of MS-SSIM. By combining these two objective functions, we show that there is a pronounced trade-off in compression quality between the human visual system and classification accuracy.