Browse › Computer Vision › Blind All-in-One Image Restoration › 3-Degradations
3-Degradations Benchmark (Blind All-in-One Image Restoration)
Blind All-in-One Image Restoration aims to remove various degradations from an input image without prior knowledge of the degradation type or severity. This task is evaluated under two setups: three-degradation and five-degradation.
Three-Degradation Setup: The objective is to restore images affected by rain, haze, and noise. Training datasets include Rain200L for deraining, RESIDE for dehazing, and WED and BSD400 for denoising with noise levels σ = 15, 25, 50. Evaluation datasets are Rain100L for deraining, SOTS (outdoor) for dehazing, and BSD68 for denoising with σ = 15, 25, 50.
Five-Degradation Setup: This setup expands to include five common image restoration tasks: rain, haze, noise, blur, and low-light conditions. Training datasets comprise Rain200L for deraining, RESIDE for dehazing, WED and BSD400 for denoising (σ = 25), GoPro for deblurring, and LoLv1 for low-light enhancement. Evaluation uses Rain100L for deraining, SOTS (outdoor) for dehazing, BSD68 for denoising (σ = 25), GoPro for deblurring, and LoLv1 for low-light enhancement.
Performance Metrics: Model performance is assessed by reporting the average PSNR and SSIM across all evaluation datasets, reflecting the overall capability to handle diverse degradations. This task challenges models to effectively restore images across multiple degradation types without specific knowledge of the degradation, emphasizing versatility and robustness in image restoration techniques.
The archive carries no text for this table; the description above is the archive's text for the task Blind All-in-One Image Restoration. archive 2025-07-28
Over time archive 2025-07-28
The chart needs JavaScript; the table below carries every value.
Direction inferred from the metric name, not from the archive: Average PSNR (higher is better), SSIM (higher is better). Points are placed at the row's paper date; 9 of 9 rows carry one.
Results archive 2025-07-28
Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.
| Paper | Code | Ran Syntology | Report | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 1 | ABAIR | 33.21 | 0.919 | – | Paper | Code | 2024 | linked, not harvested | report |
| 2 | MoCE-IR | 32.73 | 0.917 | – | Paper | – | 2024 | no code linked | report |
| 3 | HAIR | 32.70 | 0.919 | – | Paper | Code | 2024 | 13 of 18 ran · 5 unverified | report |
| 4 | AdaIR | 32.69 | 0.918 | – | Paper | Code | 2024 | 12 of 15 ran · 3 unverified | report |
| 5 | DA-RCOT | 32.60 | 0.917 | – | Paper | Code | 2024 | 6 of 6 ran · 0 unverified | report |
| 6 | RAM | 32.51 | 0.916 | – | Paper | Code | 2024 | linked, not harvested | report |
| 7 | PromptIR | 32.06 | 0.913 | – | Paper | Code | 2023 | linked, not harvested | report |
| 8 | U-WADN | 31.47 | 0.910 | – | Paper | Code | 2024 | linked, not harvested | report |
| 9 | AirNet | 31.20 | 0.910 | – | Paper | Code | 2022 | linked, not harvested | report |
All 9 rows shown. 9 link to a paper page on this site; 0 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28
Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 3 rows have a graph line, from 3 distinct papers; 3 rows (3 papers) have at least one sample that ran. Counting each paper once: Syntology ran 31 of 39 samples; 8 unverified. Separately, 24 of those 39 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections