Non-local Attention Optimized Deep Image Compression

Abstract

This paper proposes a novel Non-Local Attention Optimized Deep Image Compression (NLAIC) framework, which is built on top of the popular variational auto-encoder (VAE) structure. Our NLAIC framework embeds non-local operations in the encoders and decoders for both image and latent feature probability information (known as hyperprior) to capture both local and global correlations, and apply attention mechanism to generate masks that are used to weigh the features for the image and hyperprior, which implicitly adapt bit allocation for different features based on their importance. Furthermore, both hyperpriors and spatial-channel neighbors of the latent features are used to improve entropy coding. The proposed model outperforms the existing methods on Kodak dataset, including learned (e.g., Balle2019, Balle2018) and conventional (e.g., BPG, JPEG2000, JPEG) image compression methods, for both PSNR and MS-SSIM distortion metrics.

References in corpus (8)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Variational image compression with a scale hyperprior
• Soft-to-Hard Vector Quantization for End-to-End Learning Compressible Representations
• Joint Autoregressive and Hierarchical Priors for Learned Image Compression
• Non-Local Recurrent Network for Image Restoration
• Residual Non-local Attention Networks for Image Restoration
• Deep Image Compression via End-to-End Learning
• Gated Context Model with Embedded Priors for Deep Image Compression

Cited by in corpus (22)

• MLIC: Multi-Reference Entropy Model for Learned Image Compression
• QARV: Quantization-Aware ResNet VAE for Lossy Image Compression
• Contextformer: A Transformer with Spatio-Channel Attention for Context Modeling in Learned Image Compression
• Lossy Image Compression with Quantized Hierarchical VAEs
• Towards Robust Neural Image Compression: Adversarial Attack and Model Finetuning
• Learning Cross-Scale Weighted Prediction for Efficient Neural Video Compression
• MLIC++: Linear Complexity Multi-Reference Entropy Modeling for Learned Image Compression
• Boosting Neural Image Compression for Machines Using Latent Space Masking
• Transform Network Architectures for Deep Learning based End-to-End Image/Video Coding in Subsampled Color Spaces
• FrankenSplit: Efficient Neural Feature Compression with Shallow Variational Bottleneck Injection for Mobile Edge Computing
• High-Fidelity Variable-Rate Image Compression via Invertible Activation Transformation
• Controlling Rate, Distortion, and Realism: Towards a Single Comprehensive Neural Image Compression Model
• Efficient Visual Computing with Camera RAW Snapshots
• Generalized Gaussian Model for Learned Image Compression
• LLIC: Large Receptive Field Transform Coding with Adaptive Weights for Learned Image Compression
• Attention-Based Generative Neural Image Compression on Solar Dynamics Observatory
• Neural-based Compression Scheme for Solar Image Data
• End-to-End Optimized Image Compression with the Frequency-Oriented Transform
• L3C-Stereo: Lossless Compression for Stereo Images
• Machine Perception-Driven Image Compression: A Layered Generative Approach
• Overview of Variable Rate Coding in JPEG AI
• ANFIC: Image Compression Using Augmented Normalizing Flows