Browse State-of-the-Art › Color Image Denoising

Color Image Denoising

32 papers with code · 80 benchmarks · 9 datasets archive 2025-07-28

Computer Vision

Benchmarks archive 2025-07-28

80 leaderboard tables shown for this task, 80 with rows (a “benchmark” on this site is a table with at least one row, as on /sota), ordered by row count. “Best model” is the first row in the archive's own order at snapshot; nothing is re-ranked here and metric direction is not recorded in the archive. PwC's Trend sparklines are not in the archive, so that column is omitted. 10 shown of 80 until expanded.

Syntology 1 newer paper was judged to report on this task's benchmarks so far (Claude Sonnet 4.5, reading their tables: a model's judgement, not a result; 0 placed): CBSD68 sigma50 (1 judged, 0 placed), CBSD68 sigma15 (1 judged, 0 placed), CBSD68 sigma25 (1 judged, 0 placed), CBSD68 sigma10 (1 judged, 0 placed). Drawn from 8,886 of 18,637 newer papers checked so far (733 more had nothing to place; 41 could not be read).

DatasetBest model (first row in archive order)PaperCodeSyntologyCompare
CBSD68 sigma50 (18 rows) IPT Pre-Trained Image Processing Transformer code — Compare
CBSD68 sigma15 (10 rows) AKDT AKDT: Adaptive Kernel Dilation Transformer for Effective Image Denoising code — Compare
CBSD68 sigma25 (9 rows) AKDT AKDT: Adaptive Kernel Dilation Transformer for Effective Image Denoising code — Compare
Kodak24 sigma50 (9 rows) DMID-d(MMSE) Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Urban100 sigma50 (9 rows) Hi-IR Hierarchical Information Flow for Generalized Efficient Image Restoration — — Compare
urban100 sigma15 (8 rows) AKDT AKDT: Adaptive Kernel Dilation Transformer for Effective Image Denoising code — Compare
McMaster sigma50 (7 rows) AKDT AKDT: Adaptive Kernel Dilation Transformer for Effective Image Denoising code — Compare
CBSD68 sigma35 (6 rows) ADL Adversarial Distortion Learning for Medical Image Denoising code — Compare
Darmstadt Noise Dataset (6 rows) Image Unprocessing Unprocessing Images for Learned Raw Denoising code — Compare
Urban100 sigma25 (6 rows) Hi-IR Hierarchical Information Flow for Generalized Efficient Image Restoration — — Compare
BSD68 sigma15 (4 rows) RIDNet Real Image Denoising with Feature Attention code — Compare
BSD68 sigma25 (4 rows) RIDNet Real Image Denoising with Feature Attention code — Compare
CBSD68 sigma75 (4 rows) HyperRes Hypernetwork-Based Adaptive Image Restoration code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified Compare
McMaster sigma15 (4 rows) AKDT AKDT: Adaptive Kernel Dilation Transformer for Effective Image Denoising code — Compare
McMaster sigma25 (4 rows) AKDT AKDT: Adaptive Kernel Dilation Transformer for Effective Image Denoising code — Compare
CBSD68 sigma5 (3 rows) CBUIFD75 Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learning code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
Kodak24 sigma25 (3 rows) DMID-d Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
CBSD68 sigma10 (2 rows) CBUIFD75 Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learning code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma100 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
CBSD68 sigma150 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
CBSD68 sigma200 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
CBSD68 sigma250 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
CBSD68 sigma45 (2 rows) HyperRes Hypernetwork-Based Adaptive Image Restoration code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma55 (2 rows) HyperRes Hypernetwork-Based Adaptive Image Restoration code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma65 (2 rows) HyperRes Hypernetwork-Based Adaptive Image Restoration code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified Compare
ImageNet sigma100 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
ImageNet sigma150 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
ImageNet sigma200 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
ImageNet sigma250 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
ImageNet sigma50 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Kodak24 sigma100 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Kodak24 sigma15 (2 rows) DMID-d Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Kodak24 sigma150 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Kodak24 sigma200 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Kodak24 sigma250 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Kodak24 sigma30 (2 rows) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare
McMaster sigma100 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
McMaster sigma150 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
McMaster sigma200 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
McMaster sigma250 (2 rows) DMID-p Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
RENOIR (2 rows) BM3D RENOIR - A Dataset for Real Low-Light Image Noise Reduction code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Urban100 sigma10 (2 rows) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare
Urban100 sigma30 (2 rows) Restormer-Local Improving Image Restoration by Revisiting Global Information Aggregation code — Compare
BSD300 lambda30 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
BSD300 lambda5-50 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
BSD300 sigma25 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
BSD300 sigma5-50 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
BSD68 sigma10 (1 row) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare
BSD68 sigma30 (1 row) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare
BSD68 sigma35 (1 row) Deep CNN Denoiser Learning Deep CNN Denoiser Prior for Image Restoration code — Compare
BSD68 sigma5 (1 row) Deep CNN Denoiser Learning Deep CNN Denoiser Prior for Image Restoration code — Compare
BSD68 sigma70 (1 row) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare
BSD68 sigma75 (1 row) CSCNet Rethinking the CSC Model for Natural Images code — Compare
CBSD68 sigma20 (1 row) CBUIFD75 Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learning code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma30 (1 row) CBUIFD75 Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learning code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma40 (1 row) CBUIFD75 Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learning code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma60 (1 row) CBUIFD75 Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learning code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma70 (1 row) CBUIFD75 Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learning code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
CBSD68 sigma85 (1 row) HyperRes Hypernetwork-Based Adaptive Image Restoration code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified Compare
CellNet (1 row) DnCNN (n2t) Noise2Self: Blind Denoising by Self-Supervision code Syntology official (archive's flag): 9 ran · 19 ran (of which 0 constructed an object rather than computing a result; 16 with no instrument failure: 0 honoured, 0 violated, 16 with no contract checked; 3 where Syntology's instrument failed) · 1 unverified Compare
Hanzi (1 row) DnCNN (n2t) Noise2Self: Blind Denoising by Self-Supervision code Syntology official (archive's flag): 9 ran · 19 ran (of which 0 constructed an object rather than computing a result; 16 with no instrument failure: 0 honoured, 0 violated, 16 with no contract checked; 3 where Syntology's instrument failed) · 1 unverified Compare
Kodak sigma50 (1 row) DMID-d Stimulating Diffusion Model for Image Denoising via Adaptive... code — Compare
Kodak24 lambda30 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
Kodak24 lambda5-50 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
Kodak24 sigma10 (1 row) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare
Kodak24 sigma5-50 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
Kodak24 sigma70 (1 row) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare
Kodak25 sigma15 (1 row) FFDNet FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising code Syntology community repositories only · 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
Kodak25 sigma25 (1 row) FFDNet FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising code Syntology community repositories only · 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
Kodak25 sigma35 (1 row) FFDNet FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising code Syntology community repositories only · 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
Kodak25 sigma50 (1 row) FFDNet FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising code Syntology community repositories only · 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
Kodak25 sigma75 (1 row) FFDNet FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising code Syntology community repositories only · 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
McMaster sigma35 (1 row) FFDNet FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising code Syntology community repositories only · 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
McMaster sigma75 (1 row) FFDNet FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoising code Syntology community repositories only · 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
NTIRE 2019 Real Image Denoising Challenge (sRGB) (1 row) GRDN GRDN:Grouped Residual Dense Network for Real Image Denoising and... code — Compare
Set14 lambda30 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
Set14 lambda5-50 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
Set14 sigma25 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
Set14 sigma5-50 (1 row) SwinIA SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions — — Compare
Urban100 sigma70 (1 row) Residual Dense Network + Residual Dense Network for Image Restoration code — Compare

Syntology column: samples harvested from the paper's repositories and executed on synthesized fixtures; “ran” is not a correctness claim and does not order the table. Each line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the row's paper, the line starts with that repository's state (“official” is the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from. A dash means no Syntology record for that paper, not a recorded non-run. Each “ran” line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Libraries

Not in the archive: the export carries no per-task library table, so there is nothing to show at snapshot 2025-07-28.

Datasets archive 2025-07-28

9 datasets whose archive record lists this task, ordered by the archive's paper count.

Subtasks archive 2025-07-28

No subtask under this task in the archive's task tree.

Parent tasks archive 2025-07-28

Papers archive 2025-07-28

archive papers tagged: 51 · with a code link: 32 · where Syntology ran a sample: 11 (9 with a run with no instrument failure, 2 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (11 of 51 tagged: 9 with a run with no instrument failure, 2 where every run was a failure of Syntology's instrument)

Page 1 of 1: papers 1 to 51 of 51, in archive order: by repositories listed in the archive (most first), then newest first, not by stars (the archive holds no stars, so PwC's “Social” and “Latest” sorts cannot be reproduced). Papers that list no repository come after every paper that lists one.

Papers without a page here are shown as plain text. A Syntology line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the figure “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code); hover it for the repositories the samples that ran came from. The first sentence of the abstract is shown for the first 30 papers; every abstract is on its paper's page.

Syntology lines on 11 of the papers shown; no Syntology record for the others (a paper without an arXiv id cannot be joined to the graph, and absence from the graph layer is not a recorded non-run). “Ran” means the sample executed on a synthesized fixture, not that the paper's result was reproduced; each line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

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