Research map / Image generation / Image restoration
Priors for restoration networks
Use implicit priors, unrolling, and pretrained model priors for image restoration.
66 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Image restoration, keyword cluster “image restoration · degradation · degraded”
State History Benchmarks Tried Checklist Preprints Who Export
Where it stands
Quiet None of the tracked benchmarks has a report from 2025–26.
24 papers in 2023–24 and 10 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 7.7% to 2.3%.
No longer reported by papers here: GoPro (PSNR, last in 2024).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Image Restoration Using Very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections (NeurIPS 2016 · 1,366 citations)
- 2017 · Learning Deep CNN Denoiser Prior for Image Restoration (CVPR 2017 · 2,114 citations)
- 2018 · Deep Image Prior (CVPR 2018 · 2,004 citations)
- 2019 · Dual Residual Networks Leveraging the Potential of Paired Operations for Image Restoration (CVPR 2019 · 284 citations)
- 2020 · Bringing Old Photos Back to Life (CVPR 2020 · 239 citations)
- 2021 · Towards Real-World Blind Face Restoration With Generative Facial Prior (CVPR 2021 · 525 citations)
- 2022 · Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022 · 4,037 citations)
- 2023 · PromptIR: Prompting for All-in-One Image Restoration (NeurIPS 2023 · 85 citations)
- 2024 · Omni-Kernel Network for Image Restoration (AAAI 2024 · 275 citations)
- 2025 · DarkIR: Robust Low-Light Image Restoration (CVPR 2025 · 96 citations)
- 2026 · Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement (AAAI 2026 · 2 citations)
Earliest papers here
- Image Restoration Using Very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections (NeurIPS 2016 · 1,366 citations)
- Learning Deep CNN Denoiser Prior for Image Restoration (CVPR 2017 · 2,114 citations)
- Deeply Aggregated Alternating Minimization for Image Restoration (CVPR 2017 · 33 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
GoPro · PSNR (higher is better)
6 papers, 2022–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2022 | 32.92 | Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022 · 4,037 citations) |
| 2023 | 33.16 | IRNeXt: Rethinking Convolutional Network Design for Image Restoration (ICML 2023) |
| 2024 | 33.49 | DDR: Exploiting Deep Degradation Response as Flexible Image Descriptor (NeurIPS 2024 · 0 citations) |
What has been tried
60 papers here report 637 measured changes (a component added, removed or swapped, compared with a baseline): helped 69%, no clear effect 6%, hurt 21%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add a square large-kernel convolution to the baseline | SOTS-Indoor · PSNR | 31.32 -> 35.07 | Helped | Omni-Kernel Network for Image Restoration (AAAI 2024 · 275 citations) |
| Replace encoder EBlock with NAFBlock | LOLBlur test · PSNR | 26.51 -> 26.90 | Hurt | DarkIR: Robust Low-Light Image Restoration (CVPR 2025 · 96 citations) |
| add geometric transformations before one-direction scans | GoPro · PSNR (dB) | 33.89 -> 34.15 | Helped | Efficient Visual State Space Model for Image Deblurring (CVPR 2025 · 57 citations) |
| Use iteratively preconditioned guidance instead of pure least-squares guidance | CelebA-HQ 1K, bicubic SRx4, σe=0.05 · PSNR [dB] | 28.91 -> 29.89 | Helped | Image Restoration by Denoising Diffusion Models with Iteratively Preconditioned Guidance (CVPR 2024 · 51 citations) |
| Add PTG-SVE | LOL-real · PSNR | 24.25 -> 25.50 | Helped | Boosting Image Restoration via Priors from Pre-trained Models (CVPR 2024 · 19 citations) |
| Use a randomly initialized, fixed negative model instead of the adaptive negative model | ×4 super-resolution, average PSNR · PSNR | 28.55 -> 28.44 | Hurt | Learning from History: Task-agnostic Model Contrastive Learning for Image Restoration (AAAI 2024 · 16 citations) |
| Add latent dictionary V | CelebA-Test · LPIPS | 0.4808 -> 0.4260 | Helped | Learning Degradation-unaware Representation with Prior-based Latent Transformations for Blind Face Restoration (CVPR 2024 · 7 citations) |
| Place QUARF in the encoder rather than the decoder | Face sketch synthesis · FID | 26.89 -> 26.25 | Helped | QuARF: Quality-Adaptive Receptive Fields for Degraded Image Perception (AAAI 2025 · 1 citation) |
| Use the proposed estimated-noise injection ẑ− instead of DDIM noise ẑ | LSUN Bedroom val, σy=0.025 · PSNR [dB] | 24.64 -> 25.94 | Helped | Zero-Shot Image Restoration Using Few-Step Guidance of Consistency Models (and Beyond) (CVPR 2025 · 1 citation) |
| Increase the sliding-window size Ω | Dense haze or scenes with greater depth of view · image quality | Helped | Efficient Scene Recovery Using Luminous Flux Prior (CVPR 2024 · 1 citation) | |
| Increase sparsity from 3% to 80% | image denoising · overfitting | Hurt | Optimal Eye Surgeon: Finding image priors through sparse generators at initialization (ICML 2024 · 1 citation) | |
| Use an ensemble of restoration priors instead of a single fixed restoration prior | CS-MRI · PSNR | Helped | Stochastic Deep Restoration Priors for Imaging Inverse Problems (ICML 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 5 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- Restormer · 2 of 5 papers
- NAFNet · 2 of 5 papers
- MIMO-UNet · 2 of 5 papers
- DDRM · 2 of 5 papers
- DiffPIR · 2 of 5 papers
- DPS · 2 of 5 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 34 papers; from OpenAlex and the papers' affiliations).
- University of Science and Technology of China · 7 papers
- National University of Singapore · 2 papers
- Zhejiang University · 2 papers
- Bar-Ilan University · 2 papers
- City University of Hong Kong · 2 papers
- Peng Cheng Laboratory · 2 papers
- Nanjing University of Science and Technology · 2 papers
- Google · 2 papers
- Nanyang Technological University · 2 papers
Use it
Download BibTeX of 38 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Efficient restoration architectures (113 papers)
- Diffusion-based universal restoration (61 papers)
- Real-world restoration with unaligned data (40 papers)
Elsewhere in the topic
- Dehazing and deraining (118 papers)
- Inverse imaging with learned priors (109 papers)
- Image harmonization and color transformation (102 papers)
- Reflection removal (95 papers)
- Pansharpening and spectral fusion (84 papers)
- Image deblurring (73 papers)
- Low-light image enhancement (72 papers)
- Learned image compression (69 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).