Universal End-to-End Neural Network for Lossy Image Compression

TL;DR

This paper introduces a universal neural network-based lossy image compression method that adjusts image quality during inference, demonstrating high efficiency and broad applicability across architectures and training methods.

Contribution

It proposes a simple, inference-only input scaling technique for variable bitrate image compression, enhancing universality and reducing complexity without changing model architecture or loss functions.

Findings

Effective across diverse architectures and training methods

Reduces computational complexity and memory requirements

Shows promising results in variable-rate image compression

Abstract

This paper presents variable bitrate lossy image compression using a VAE-based neural network. An adaptable image quality adjustment strategy is proposed. The key innovation involves adeptly adjusting the input scale exclusively during the inference process, resulting in an exceptionally efficient rate-distortion mechanism. Through extensive experimentation, across diverse VAE-based compression architectures (CNN, ViT) and training methodologies (MSE, SSIM), our approach exhibits remarkable universality. This success is attributed to the inherent generalization capacity of neural networks. Unlike methods that adjust model architecture or loss functions, our approach emphasizes simplicity, reducing computational complexity and memory requirements. The experiments not only highlight the effectiveness of our approach but also indicate its potential to drive advancements in variable-rate neural network lossy image compression methodologies.