Deep Learning-Driven Ultra-High-Definition Image Restoration: A Survey

November 13, 2025 · View on GitHub

Deep Learning-Driven Ultra-High-Definition Image Restoration: A Survey


Overview

Datasets

Datasets Restoration Task Number(Train/Test) Synthetic/Real Format Publication DownLoad
UHDSR4K Image Super-Resolution 5,999 / 2,100 Real PNG ICCV'21 Link
4KRD Video Deblurring 19,958 / 1600 Synthetic JPG ICCV'21 Link
UHD-Blur Image Deblurring 1,964 / 300 Synthetic JPG AAAI'24 Link
4KID Image Dehazing 8,000 / 200 Synthetic JPG CVPR'21 Link
Updated 4KID Image Dehazing 10,400 / 200 Synthetic JPG Signal Process.'24
UHD-Haze Image Dehazing 2,290 / 230 Synthetic JPG AAAI'24 Link
4KIL Low-Light Image Enhancing 1,000 / 100 Synthetic JPG Inf. Sci'22
UHD-LOL4K Low-Light Image Enhancing 5,999 / 2,100 Synthetic PNG AAAI'23 Link
UHD-LL Low-Light Image Enhancing 2,000 / 150 Real JPG ICLR'23 Link
4K-Rain13K Image Deraining 12,500 / 500 Synthetic JPG ArXiv'24 Link
4K-RealRain Image Deraining - / 320 Real JPG ArXiv'24 Link
UHD-Rain Image Deraining 3,000 / 200 Synthetic JPG TCSVT'25 Link
UHD-Snow Image Deraining 3,000 / 200 Synthetic JPG TCSVT'25 Link

UHD Image Restoration Methods

Category
(分类)
Methods
(方法)
Tasks
(任务)
Title
(标题)
Venue
(发表场所)
Source
(资源)
Downsampling-Enhancement-Upsampling MANet SR Benchmarking ultra-high-definition image superresolution ICCV'21 Paper/
Downsampling-Enhancement-Upsampling UHD Dehaze Ultra-high-definition image dehazing via multi-guided bilateral learning CVPR'21 Paper/Code
Downsampling-Enhancement-Upsampling Lin et al. LLIE UHD low-light image enhancement via interpretable bilateral learning Inf. Sci'22 Paper/Code
Downsampling-Enhancement-Upsampling Cubic-Mixer Deblur Ultra-high-definition image deblurring via multi-scale cubic-mixer ArXiv'24 Paper/Code
Downsampling-Enhancement-Upsampling UHDFour LLIE Embedding fourier for ultra-high-definition low-light image enhancement ICLR'23 Paper/Code
Downsampling-Enhancement-Upsampling DMixer Dehaze Dimensional transformation mixer for ultra-high-definition industrial camera dehazing IEEE Trans. Ind. Inf.'24 Paper/
Downsampling-Enhancement-Upsampling MixNet LLIE/Deblur Mixnet: Towards effective and efficient UHD low-light image enhancement ArXiv'24 Paper/>
Downsampling-Enhancement-Upsampling UDR-Mixer Derain Towards ultra-high-definition image deraining: A benchmark and an efficient method ArXiv'24 Paper/Code
Downsampling-Enhancement-Upsampling UHDformer LLIE/Deblur/Dehaze Correlation matching transformation transformers for UHD image restoration AAAI'24 Paper/Code
Downsampling-Enhancement-Upsampling UHDDIP LLIE/Deblur/Dehaze/
Derain/Desnow
Ultra-high-definition restoration: New benchmarks and A dual interaction prior-driven solution TCSVT'25 Paper/Code
Downsampling-Enhancement-Upsampling SimpleIR All-in-One Review learning: Advancing all-in-one ultra-high-definition image restoration training method ArXiv'24 Paper/
Encoder-Decoder with Stepwise Up-downsampling UHDVD Deblur Review learning: Advancing all-in-one ultra-high-definition image restoration training method ICCV'21 Paper/Code
Encoder-Decoder with Stepwise Up-downsampling LLFormer Deblur Ultrahigh-definition low-light image enhancement: A benchmark and transformer-based method AAAI'23 Paper/Code
Encoder-Decoder with Stepwise Up-downsampling LapDehazeNet Dehaze Single UHD image dehazing via interpretable pyramid network Signal Process.'24 Paper/Code
Encoder-Decoder with Stepwise Up-downsampling Wave-Mamba LLIE Wave-mamba:Wavelet state space model for ultra-high-definition low-light image enhancement ACM MM'24 Paper/Code
Encoder-Decoder with Stepwise Up-downsampling TSFormer LLIE/Deblur/Dehaze/
Derain/Desnow
Tsformer: A robust framework for efficient uhd image restoration ArXiv'24 Paper/
Encoder-Decoder with Stepwise Up-downsampling FEVAE-UHD LLIE/Deblur/Dehaze DreamUHD: Frequency Enhanced Variational Autoencoder for Ultra-High-Definition Image Restoration AAAI'25 Paper/Code
Encoder-Decoder with Stepwise Up-downsampling UHD-processor All-in-One UHD-processer: Unified UHD Image Restoration with Progressive Frequency Learning and Degradation-aware Prompts CVPR'25 Paper/
Encoder-Decoder with Stepwise Up-downsampling ERR LLIE/Deblur/Dehaze/
Derain
From Zero to Detail: Deconstructing Ultra-High-Definition Image Restoration from Progressive Spectral Perspective CVPR'25 Paper/Code
Resampling-Enhancement NSEN LLIE/Dehaze Learning non-uniform-sampling for ultra-high-definition image enhancement ACM MM'23 Paper/
Resampling-Enhancement LMAR LLIE Empowering resampling operation for ultra-high-definition image enhancement with model-aware guidance CVPR'24 Paper/Code