On Efficient Neural Network Architectures for Image Compression

TL;DR

This paper investigates how different neural network designs affect the efficiency and performance of image compression models, proposing new models that balance quality and computational cost.

Contribution

It empirically evaluates various network architectures and introduces efficient models that achieve comparable performance to state-of-the-art methods with lower complexity.

Findings

Efficient models match recent best-performing methods in quality.

Hierarchical and space-channel context models improve compression.

Significant reduction in computational complexity achieved.

Abstract

Recent advances in learning-based image compression typically come at the cost of high complexity. Designing computationally efficient architectures remains an open challenge. In this paper, we empirically investigate the impact of different network designs in terms of rate-distortion performance and computational complexity. Our experiments involve testing various transforms, including convolutional neural networks and transformers, as well as various context models, including hierarchical, channel-wise, and space-channel context models. Based on the results, we present a series of efficient models, the final model of which has comparable performance to recent best-performing methods but with significantly lower complexity. Extensive experiments provide insights into the design of architectures for learned image compression and potential direction for future research. The code is available at \url{https://gitlab.com/viper-purdue/efficient-compression}.