Pyramid Attention Network for Image Restoration

Abstract Self-similarity refers to the image prior widely used in restoration algorithms that small but similar patterns tend occur at different locations and scales. However, recent advanced deep convolutional neural network-based methods for do not take full advantage of self-similarities by relying on self-attention modules only process information same scale. To solve this problem, we present a novel Pyramid Attention module restoration, which captures long-range feature correspondences from multi-scale pyramid. Inspired fact corruptions, such as noise or compression artifacts, drop drastically coarser scales, our attention is designed be able borrow clean signals their “clean” levels. The proposed pyramid generic building block can flexibly integrated into various architectures. Its effectiveness validated through extensive experiments multiple tasks: denoising, demosaicing, artifact reduction, super resolution. Without any bells whistles, PANet (pyramid with simple network backbones) produce state-of-the-art results superior accuracy visual quality. Our code available https://github.com/SHI-Labs/Pyramid-Attention-Networks