Slimmable Compressive Autoencoders for Practical Neural Image Compression

Abstract

Neural image compression leverages deep neural networks to outperform traditional image codecs in rate-distortion performance. However, the resulting models are also heavy, computationally demanding and generally optimized for a single rate, limiting their practical use. Focusing on practical image compression, we propose slimmable compressive autoencoders (SlimCAEs), where rate (R) and distortion (D) are jointly optimized for different capacities. Once trained, encoders and decoders can be executed at different capacities, leading to different rates and complexities. We show that a successful implementation of SlimCAEs requires suitable capacity-specific RD tradeoffs. Our experiments show that SlimCAEs are highly flexible models that provide excellent rate-distortion performance, variable rate, and dynamic adjustment of memory, computational cost and latency, thus addressing the main requirements of practical image compression.

Accepted to CVPR 2021

References in corpus (8)

• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Neural Architecture Search with Reinforcement Learning
• Designing Neural Network Architectures using Reinforcement Learning
• Soft-to-Hard Vector Quantization for End-to-End Learning Compressible Representations
• Joint Autoregressive and Hierarchical Priors for Learned Image Compression
• Context-adaptive Entropy Model for End-to-end Optimized Image Compression
• Lossy Image Compression with Compressive Autoencoders
• Computationally Efficient Neural Image Compression