Variable Rate Deep Image Compression With a Conditional Autoencoder

TL;DR

This paper introduces a single, flexible deep image compression model using a conditional autoencoder that efficiently adapts to different rates with improved rate-distortion performance.

Contribution

The authors develop a unified variable-rate image compression framework with a conditional autoencoder, eliminating the need for multiple trained models for different rates.

Findings

Outperforms traditional codecs like JPEG2000 and BPG in rate-distortion trade-off.

Achieves comparable or better results than state-of-the-art learned compression models.

Provides effective coarse and fine rate control via Lagrange multiplier and quantization bin size.

Abstract

In this paper, we propose a novel variable-rate learned image compression framework with a conditional autoencoder. Previous learning-based image compression methods mostly require training separate networks for different compression rates so they can yield compressed images of varying quality. In contrast, we train and deploy only one variable-rate image compression network implemented with a conditional autoencoder. We provide two rate control parameters, i.e., the Lagrange multiplier and the quantization bin size, which are given as conditioning variables to the network. Coarse rate adaptation to a target is performed by changing the Lagrange multiplier, while the rate can be further fine-tuned by adjusting the bin size used in quantizing the encoded representation. Our experimental results show that the proposed scheme provides a better rate-distortion trade-off than the traditional variable-rate image compression codecs such as JPEG2000 and BPG. Our model also shows comparable and sometimes better performance than the state-of-the-art learned image compression models that deploy multiple networks trained for varying rates.