TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Color Image Denoising	CBSD68 sigma15	HyperRes	PSNR	34.02	# 4
Color Image Denoising	CBSD68 sigma15	HyperRes	SSIM	0.95	# 1
Color Image Denoising	CBSD68 sigma25	HyperRes	PSNR	31.41	# 3
Color Image Denoising	CBSD68 sigma25	HyperRes	SSIM	0.89	# 1
Color Image Denoising	CBSD68 sigma35	HyperRes	PSNR	29.8	# 2
Color Image Denoising	CBSD68 sigma35	HyperRes	SSIM	0.85	# 1
Color Image Denoising	CBSD68 sigma45	HyperRes	PSNR	28.64	# 1
Color Image Denoising	CBSD68 sigma45	HyperRes	SSIM	0.81	# 1
Color Image Denoising	CBSD68 sigma5	HyperRes	PSNR	39.96	# 2
Color Image Denoising	CBSD68 sigma5	HyperRes	SSIM	0.98	# 1
Color Image Denoising	CBSD68 sigma55	HyperRes	PSNR	27.78	# 1
Color Image Denoising	CBSD68 sigma55	HyperRes	SSIM	0.78	# 1
Color Image Denoising	CBSD68 sigma65	HyperRes	PSNR	27.06	# 1
Color Image Denoising	CBSD68 sigma65	HyperRes	SSIM	0.76	# 1
Color Image Denoising	CBSD68 sigma75	HyperRes	PSNR	26.46	# 1
Color Image Denoising	CBSD68 sigma75	HyperRes	SSIM	0.73	# 1
Color Image Denoising	CBSD68 sigma85	HyperRes	PSNR	25.95	# 1
Color Image Denoising	CBSD68 sigma85	HyperRes	SSIM	0.71	# 1
JPEG Artifact Correction	Live1 (Quality 10 Grayscale)	HyperRes	PSNR	28.81	# 12
JPEG Artifact Correction	Live1 (Quality 10 Grayscale)	HyperRes	SSIM	0.82	# 10
JPEG Artifact Correction	LIVE1 (Quality 30 Grayscale)	HyperRes	PSNR	32.56	# 6
JPEG Artifact Correction	LIVE1 (Quality 30 Grayscale)	HyperRes	SSIM	0.91	# 4
JPEG Artifact Correction	Live1 (Quality 50 Grayscale)	HyperRes	PSNR	34.39	# 1
JPEG Artifact Correction	Live1 (Quality 50 Grayscale)	HyperRes	SSIM	0.94	# 1
JPEG Artifact Correction	Live1 (Quality 70 Grayscale)	HyperRes	PSNR	36.38	# 1
JPEG Artifact Correction	Live1 (Quality 70 Grayscale)	HyperRes	SSIM	0.96	# 1
JPEG Artifact Correction	Live1 (Quality 80 Grayscale)	HyperRes	PSNR	38.09	# 1
JPEG Artifact Correction	Live1 (Quality 80 Grayscale)	HyperRes	SSIM	0.97	# 1
Image Super-Resolution	Set5 - 2x upscaling	HyperRes	PSNR	36.69	# 30
Image Super-Resolution	Set5 - 2x upscaling	HyperRes	SSIM	0.94	# 17
Image Super-Resolution	Set5 - 3x upscaling	HyperRes	PSNR	29.77	# 24
Image Super-Resolution	Set5 - 3x upscaling	HyperRes	SSIM	0.84	# 12
Image Super-Resolution	Set5 - 5x upscaling	HyperRes	PSNR	25.63	# 1
Image Super-Resolution	Set5 - 5x upscaling	HyperRes	SSIM	0.74	# 1
Image Super-Resolution	Set5 - 6x upscaling	HyperRes	PSNR	24.92	# 1
Image Super-Resolution	Set5 - 6x upscaling	HyperRes	SSIM	0.71	# 1

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma45)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma45?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma55)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma55?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma65)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma65?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma75)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma75?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma85)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma85?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/jpeg-artifact-correction-on-live1-quality-50)](https://paperswithcode.com/sota/jpeg-artifact-correction-on-live1-quality-50?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/jpeg-artifact-correction-on-live1-quality-70)](https://paperswithcode.com/sota/jpeg-artifact-correction-on-live1-quality-70?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/jpeg-artifact-correction-on-live1-quality-80)](https://paperswithcode.com/sota/jpeg-artifact-correction-on-live1-quality-80?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/image-super-resolution-on-set5-5x-upscaling)](https://paperswithcode.com/sota/image-super-resolution-on-set5-5x-upscaling?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/image-super-resolution-on-set5-6x-upscaling)](https://paperswithcode.com/sota/image-super-resolution-on-set5-6x-upscaling?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma35)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma35?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma5)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma5?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma25)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma25?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/color-image-denoising-on-cbsd68-sigma15)](https://paperswithcode.com/sota/color-image-denoising-on-cbsd68-sigma15?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/jpeg-artifact-correction-on-live1-quality-30-1)](https://paperswithcode.com/sota/jpeg-artifact-correction-on-live1-quality-30-1?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/jpeg-artifact-correction-on-live1-quality-10-1)](https://paperswithcode.com/sota/jpeg-artifact-correction-on-live1-quality-10-1?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/image-super-resolution-on-set5-3x-upscaling)](https://paperswithcode.com/sota/image-super-resolution-on-set5-3x-upscaling?p=one-size-fits-all-hypernetwork-for-tunable)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/one-size-fits-all-hypernetwork-for-tunable/image-super-resolution-on-set5-2x-upscaling)](https://paperswithcode.com/sota/image-super-resolution-on-set5-2x-upscaling?p=one-size-fits-all-hypernetwork-for-tunable)`

Hypernetwork-Based Adaptive Image Restoration

13 Jun 2022 · Shai Aharon, Gil Ben-Artzi ·

Adaptive image restoration models can restore images with different degradation levels at inference time without the need to retrain the model. We present an approach that is highly accurate and allows a significant reduction in the number of parameters. In contrast to existing methods, our approach can restore images using a single fixed-size model, regardless of the number of degradation levels. On popular datasets, our approach yields state-of-the-art results in terms of size and accuracy for a variety of image restoration tasks, including denoising, deJPEG, and super-resolution.

PDF Abstract

Code

Add Remove Mark official

ifryed/HyperRes official

↳ Quickstart in

Colab

Tasks

Add Remove

Color Image Denoising

Denoising

Image Denoising

Image Restoration

Image Super-Resolution

JPEG Artifact Correction

JPEG Artifact Removal

Super-Resolution

Datasets

DIV2K

Set5 CBSD68

LIVE (Public-Domain Subjective Image Quality Database)

Results from the Paper

Edit

Ranked #1 on Color Image Denoising on CBSD68 sigma55

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Color Image Denoising	CBSD68 sigma15	HyperRes	PSNR	34.02	# 4	Compare
Color Image Denoising	CBSD68 sigma15	HyperRes	SSIM	0.95	# 1	Compare
Color Image Denoising	CBSD68 sigma25	HyperRes	PSNR	31.41	# 3	Compare
Color Image Denoising	CBSD68 sigma25	HyperRes	SSIM	0.89	# 1	Compare
Color Image Denoising	CBSD68 sigma35	HyperRes	PSNR	29.8	# 2	Compare
Color Image Denoising	CBSD68 sigma35	HyperRes	SSIM	0.85	# 1	Compare
Color Image Denoising	CBSD68 sigma45	HyperRes	PSNR	28.64	# 1	Compare
Color Image Denoising	CBSD68 sigma45	HyperRes	SSIM	0.81	# 1	Compare
Color Image Denoising	CBSD68 sigma5	HyperRes	PSNR	39.96	# 2	Compare
Color Image Denoising	CBSD68 sigma5	HyperRes	SSIM	0.98	# 1	Compare
Color Image Denoising	CBSD68 sigma55	HyperRes	PSNR	27.78	# 1	Compare
Color Image Denoising	CBSD68 sigma55	HyperRes	SSIM	0.78	# 1	Compare
Color Image Denoising	CBSD68 sigma65	HyperRes	PSNR	27.06	# 1	Compare
Color Image Denoising	CBSD68 sigma65	HyperRes	SSIM	0.76	# 1	Compare
Color Image Denoising	CBSD68 sigma75	HyperRes	PSNR	26.46	# 1	Compare
Color Image Denoising	CBSD68 sigma75	HyperRes	SSIM	0.73	# 1	Compare
Color Image Denoising	CBSD68 sigma85	HyperRes	PSNR	25.95	# 1	Compare
Color Image Denoising	CBSD68 sigma85	HyperRes	SSIM	0.71	# 1	Compare
JPEG Artifact Correction	Live1 (Quality 10 Grayscale)	HyperRes	PSNR	28.81	# 12	Compare
JPEG Artifact Correction	Live1 (Quality 10 Grayscale)	HyperRes	SSIM	0.82	# 10	Compare
JPEG Artifact Correction	LIVE1 (Quality 30 Grayscale)	HyperRes	PSNR	32.56	# 6	Compare
JPEG Artifact Correction	LIVE1 (Quality 30 Grayscale)	HyperRes	SSIM	0.91	# 4	Compare
JPEG Artifact Correction	Live1 (Quality 50 Grayscale)	HyperRes	PSNR	34.39	# 1	Compare
JPEG Artifact Correction	Live1 (Quality 50 Grayscale)	HyperRes	SSIM	0.94	# 1	Compare
JPEG Artifact Correction	Live1 (Quality 70 Grayscale)	HyperRes	PSNR	36.38	# 1	Compare
JPEG Artifact Correction	Live1 (Quality 70 Grayscale)	HyperRes	SSIM	0.96	# 1	Compare
JPEG Artifact Correction	Live1 (Quality 80 Grayscale)	HyperRes	PSNR	38.09	# 1	Compare
JPEG Artifact Correction	Live1 (Quality 80 Grayscale)	HyperRes	SSIM	0.97	# 1	Compare
Image Super-Resolution	Set5 - 2x upscaling	HyperRes	PSNR	36.69	# 30	Compare
Image Super-Resolution	Set5 - 2x upscaling	HyperRes	SSIM	0.94	# 17	Compare
Image Super-Resolution	Set5 - 3x upscaling	HyperRes	PSNR	29.77	# 24	Compare
Image Super-Resolution	Set5 - 3x upscaling	HyperRes	SSIM	0.84	# 12	Compare
Image Super-Resolution	Set5 - 5x upscaling	HyperRes	PSNR	25.63	# 1	Compare
Image Super-Resolution	Set5 - 5x upscaling	HyperRes	SSIM	0.74	# 1	Compare
Image Super-Resolution	Set5 - 6x upscaling	HyperRes	PSNR	24.92	# 1	Compare
Image Super-Resolution	Set5 - 6x upscaling	HyperRes	SSIM	0.71	# 1	Compare

Methods

Add Remove

Batch Normalization • HyperNetwork • ReLU • Residual Block • Residual Connection • ResNet • Tunable Network

Edit Social Preview

Hypernetwork-Based Adaptive Image Restoration

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Datasets Edit

Results from the Paper Edit

Methods Edit Add Remove

Code

Add Remove Mark official

Tasks

Add Remove

Datasets

Results from the Paper

Edit

Methods

Add Remove