# Neural ISP on Nighttime raw-sRGB noise-free (test)

Source: [Neural ISP on Nighttime raw-sRGB noise-free (test)](<https://www.sota2.com/research/sota/neural-isp-on-nighttime-raw-srgb-noise-free-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:55:36.278Z

Evaluation of Neural Image Signal Processing (ISP) performance on the test split of the Nighttime raw-sRGB noise-free dataset, measured using PSNR, SSIM, and Delta E.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Neural ISP |
| Dataset (including any stated split) | Nighttime raw-sRGB noise-free (test) |
| Metric | PSNR |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [PSNR](<https://www.sota2.com/research/sota/neural-isp-on-nighttime-raw-srgb-noise-free-test.md?metric=PSNR>): Higher is better; unit: SCALAR.

- [SSIM](<https://www.sota2.com/research/sota/neural-isp-on-nighttime-raw-srgb-noise-free-test.md?metric=SSIM>): Higher is better; unit: SCALAR.

- [Delta E](<https://www.sota2.com/research/sota/neural-isp-on-nighttime-raw-srgb-noise-free-test.md?metric=Delta%20E>): Lower is better; unit: SCALAR.

## Recorded results

| Method | PSNR | Result date | Report paper |
| --- | --- | --- | --- |
| Supervised | 46.19 | Not recorded | 68fa6cb31e209131cea3554d |
| Ours 90% + Real 10% | 46.01 | Not recorded | 68fa6cb31e209131cea3554d |
| Ours 95% + Real 5% | 45.99 | Not recorded | 68fa6cb31e209131cea3554d |
| Ours | 45.28 | Not recorded | 68fa6cb31e209131cea3554d |
| Global relight | 43.56 | Not recorded | 68fa6cb31e209131cea3554d |
| Day dimmed | 42.39 | Not recorded | 68fa6cb31e209131cea3554d |
| Day | 41.16 | Not recorded | 68fa6cb31e209131cea3554d |
| CycleGAN | 40.01 | Not recorded | 68fa6cb31e209131cea3554d |

### Supervised

PSNR: 46.19. Result ID: 698f2f503a6c314849f6cc93.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa644b1e209131cea03c6a.md>)

Recorded evaluation configuration: {"Training data partition":"Purely real","Evaluation protocol":"Three-fold cross validation"}. This record does not assert independent reproduction.

### Ours 90% + Real 10%

PSNR: 46.01. Result ID: 698f2f503a6c314849f6cc92.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Implementation](<https://github.com/SamsungLabs/day-to-night>)

Recorded evaluation configuration: {"Training data partition":"Mix (synthetic and real)","Data composition":"90% Synthetic + 10% Real"}. This record does not assert independent reproduction.

### Ours 95% + Real 5%

PSNR: 45.99. Result ID: 698f2f503a6c314849f6cc91.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Implementation](<https://github.com/SamsungLabs/day-to-night>)

Recorded evaluation configuration: {"Training data partition":"Mix (synthetic and real)","Data composition":"95% Synthetic + 5% Real"}. This record does not assert independent reproduction.

### Ours

PSNR: 45.28. Result ID: 698f2f503a6c314849f6cc90.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Implementation](<https://github.com/SamsungLabs/day-to-night>)

Recorded evaluation configuration: {"Training data partition":"Synthetic only","Data source":"Proposed day-to-night synthesis"}. This record does not assert independent reproduction.

### Global relight

PSNR: 43.56. Result ID: 698f2f503a6c314849f6cc8e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

Recorded evaluation configuration: {"Training data partition":"Synthetic only","Data source":"Global relight"}. This record does not assert independent reproduction.

### Day dimmed

PSNR: 42.39. Result ID: 698f2f503a6c314849f6cc8d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

Recorded evaluation configuration: {"Training data partition":"Synthetic only","Data source":"Day dimmed images"}. This record does not assert independent reproduction.

### Day

PSNR: 41.16. Result ID: 698f2f503a6c314849f6cc8c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

Recorded evaluation configuration: {"Training data partition":"Synthetic only","Data source":"Day images"}. This record does not assert independent reproduction.

### CycleGAN

PSNR: 40.01. Result ID: 698f2f503a6c314849f6cc8f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3554d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68351e209131cea1ac7a.md>)

Recorded evaluation configuration: {"Training data partition":"Synthetic only","Data source":"CycleGAN synthesis"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.