HINet: Half Instance Normalization Network for Image Restoration (2021-05-13T00:00:00.000000Z)

TL;DR

A novel block is presented: Half Instance Normalization Block (HIN Block), to boost the performance of image restoration networks and a simple and powerful multi-stage network named HINet is designed, which surpasses the state-of-the-art (SOTA) on various image restoration tasks.

Abstract

In this paper, we explore the role of Instance Normalization in low-level vision tasks. Specifically, we present a novel block: Half Instance Normalization Block (HIN Block), to boost the performance of image restoration networks. Based on HIN Block, we design a simple and powerful multi-stage network named HINet, which consists of two subnetworks. With the help of HIN Block, HINet surpasses the state-of-the-art (SOTA) on various image restoration tasks. For image denoising, we exceed it 0.11dB and 0.28 dB in PSNR on SIDD dataset, with only 7.5% and 30% of its multiplier-accumulator operations (MACs), 6.8× and 2.9× speedup respectively. For image deblurring, we get comparable performance with 22.5% of its MACs and 3.3× speedup on REDS and GoPro datasets. For image deraining, we exceed it by 0.3 dB in PSNR on the average result of multiple datasets with 1.4× speedup. With HINet, we won the 1st place on the NTIRE 2021 Image Deblurring Challenge - Track2. JPEG Artifacts, with a PSNR of 29.70.

Authors

Liangyu Chen

3 papers

J. Zhang

3 papers

Xin Lu

1 papers

TL;DR

Abstract

Authors

References63 items

NTIRE 2021 Challenge on Image Deblurring

Multi-Stage Progressive Image Restoration

Practical Deep Raw Image Denoising on Mobile Devices

Dual Adversarial Network: Toward Real-world Noise Removal and Noise Generation

End-to-End Object Detection with Transformers

Spatially-Attentive Patch-Hierarchical Network for Adaptive Motion Deblurring

Deblurring by Realistic Blurring

Multi-Scale Progressive Fusion Network for Single Image Deraining

CycleISP: Real Image Restoration via Improved Data Synthesis

Transfer Learning From Synthetic to Real-Noise Denoising With Adaptive Instance Normalization

Spatial-Adaptive Network for Single Image Denoising

Multi-Temporal Recurrent Neural Networks For Progressive Non-Uniform Single Image Deblurring With Incremental Temporal Training

Variational Denoising Network: Toward Blind Noise Modeling and Removal

DeblurGAN-v2: Deblurring (Orders-of-Magnitude) Faster and Better

Densely Residual Laplacian Super-Resolution

NTIRE 2019 Challenge on Video Deblurring and Super-Resolution: Dataset and Study

Dynamic Scene Deblurring With Parameter Selective Sharing and Nested Skip Connections

Uncertainty Guided Multi-Scale Residual Learning-Using a Cycle Spinning CNN for Single Image De-Raining

EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

Real Image Denoising With Feature Attention

Deep Stacked Hierarchical Multi-Patch Network for Image Deblurring

FCOS: Fully Convolutional One-Stage Object Detection

Progressive Image Deraining Networks: A Better and Simpler Baseline

Wide Activation for Efficient and Accurate Image Super-Resolution

ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design

Semi-Supervised Transfer Learning for Image Rain Removal

Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net

Recurrent Squeeze-and-Excitation Context Aggregation Net for Single Image Deraining

Toward Convolutional Blind Denoising of Real Photographs

A High-Quality Denoising Dataset for Smartphone Cameras

Dynamic Scene Deblurring Using Spatially Variant Recurrent Neural Networks

Lightweight Pyramid Networks for Image Deraining

Context Encoding for Semantic Segmentation

Group Normalization

Density-Aware Single Image De-raining Using a Multi-stream Dense Network

Scale-Recurrent Network for Deep Image Deblurring

DeblurGAN: Blind Motion Deblurring Using Conditional Adversarial Networks

Enhanced Deep Residual Networks for Single Image Super-Resolution

ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices

Removing Rain from Single Images via a Deep Detail Network

Arbitrary Style Transfer in Real-Time with Adaptive Instance Normalization

Image De-Raining Using a Conditional Generative Adversarial Network

Improved Texture Networks: Maximizing Quality and Diversity in Feed-Forward Stylization and Texture Synthesis

From Motion Blur to Motion Flow: A Deep Learning Solution for Removing Heterogeneous Motion Blur

Deep Multi-scale Convolutional Neural Network for Dynamic Scene Deblurring

Deep Joint Rain Detection and Removal from a Single Image

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

Clearing the Skies: A Deep Network Architecture for Single-Image Rain Removal

Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

SGDR: Stochastic Gradient Descent with Warm Restarts

Instance Normalization: The Missing Ingredient for Fast Stylization

Layer Normalization

Rain Streak Removal Using Layer Priors

Texture Networks: Feed-forward Synthesis of Textures and Stylized Images

Deep Residual Learning for Image Recognition

U-Net: Convolutional Networks for Biomedical Image Segmentation

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Dynamic Scene Deblurring

Unnatural L0 Sparse Representation for Natural Image Deblurring

Image denoising: Can plain neural networks compete with BM3D?

Non-uniform Deblurring for Shaken Images

Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering

Rectifier Nonlinearities Improve Neural Network Acoustic Models

Field of Study

Journal Information

Name

Page