# Spatial Aesthetic Image Quality Assessment on SA-BENCH

Source: [Spatial Aesthetic Image Quality Assessment on SA-BENCH](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench>)

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

Data updated: 2026-02-26T05:48:33.098Z

Evaluation of models on assessing the spatial aesthetic quality of images using the SA-BENCH dataset, typically measured using PLCC and SRCC across layout, harmony, lighting, and distortion dimensions.

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 | Spatial Aesthetic Image Quality Assessment |
| Dataset (including any stated split) | SA-BENCH |
| Metric | Layout PLCC |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 25 of 25 |
| 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

- [Layout PLCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Layout%20PLCC>): Higher is better; unit: SCALAR.

- [Layout SRCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Layout%20SRCC>): Higher is better; unit: SCALAR.

- [Harmony PLCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Harmony%20PLCC>): Higher is better; unit: SCALAR.

- [Harmony SRCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Harmony%20SRCC>): Higher is better; unit: SCALAR.

- [Lighting PLCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Lighting%20PLCC>): Higher is better; unit: SCALAR.

- [Lighting SRCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Lighting%20SRCC>): Higher is better; unit: SCALAR.

- [Distortion PLCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Distortion%20PLCC>): Higher is better; unit: SCALAR.

- [Distortion SRCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Distortion%20SRCC>): Higher is better; unit: SCALAR.

- [Overall PLCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Overall%20PLCC>): Higher is better; unit: SCALAR.

- [Overall SRCC](<https://www.sota2.com/research/sota/spatial-aesthetic-image-quality-assessment-on-sa-bench.md?metric=Overall%20SRCC>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Layout PLCC | Result date | Report paper |
| --- | --- | --- | --- |
| GLM-4-1V-9B | 0.831 | Not recorded | 6932675cd6b0dd3342934591 |
| SA-IQA | 0.831 | Not recorded | 6932675cd6b0dd3342934591 |
| Qwen3-VL-4B | 0.821 | Not recorded | 6932675cd6b0dd3342934591 |
| Qwen3-VL-8B | 0.814 | Not recorded | 6932675cd6b0dd3342934591 |
| Ovis2-5-2B | 0.812 | Not recorded | 6932675cd6b0dd3342934591 |
| Qwen3-VL-2B | 0.808 | Not recorded | 6932675cd6b0dd3342934591 |
| mPLUG-Owl3-7B | 0.784 | Not recorded | 6932675cd6b0dd3342934591 |
| InternVL3-5-8B | 0.774 | Not recorded | 6932675cd6b0dd3342934591 |
| Q-Eval-Score | 0.265 | Not recorded | 6932675cd6b0dd3342934591 |
| gemini-2.5-pro | 0.246 | Not recorded | 6932675cd6b0dd3342934591 |
| DeQA-Score | 0.201 | Not recorded | 6932675cd6b0dd3342934591 |
| CLIPIQA | 0.197 | Not recorded | 6932675cd6b0dd3342934591 |
| MANIQA | 0.194 | Not recorded | 6932675cd6b0dd3342934591 |
| gpt-5 | 0.185 | Not recorded | 6932675cd6b0dd3342934591 |
| HyperIQA | 0.179 | Not recorded | 6932675cd6b0dd3342934591 |
| DBCNN | 0.162 | Not recorded | 6932675cd6b0dd3342934591 |
| Compare2Score | 0.16 | Not recorded | 6932675cd6b0dd3342934591 |
| Q-SiT | 0.141 | Not recorded | 6932675cd6b0dd3342934591 |
| MUSIQ | 0.121 | Not recorded | 6932675cd6b0dd3342934591 |
| qwen-vl-max | 0.088 | Not recorded | 6932675cd6b0dd3342934591 |
| Q-Align-aesthetics | 0.075 | Not recorded | 6932675cd6b0dd3342934591 |
| Q-Align-quality | 0.057 | Not recorded | 6932675cd6b0dd3342934591 |
| Q-Insight | 0.053 | Not recorded | 6932675cd6b0dd3342934591 |
| gpt-4o | 0.017 | Not recorded | 6932675cd6b0dd3342934591 |
| claude3.5-sonnet | -0.001 | Not recorded | 6932675cd6b0dd3342934591 |

### GLM-4-1V-9B

Layout PLCC: 0.831. Result ID: 69988ccb5698665ec4ee5de9.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74521e209131cea610b6.md>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### SA-IQA

Layout PLCC: 0.831. Result ID: 69988ccb5698665ec4ee5dee.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### Qwen3-VL-4B

Layout PLCC: 0.821. Result ID: 69988ccb5698665ec4ee5dec.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### Qwen3-VL-8B

Layout PLCC: 0.814. Result ID: 69988ccb5698665ec4ee5ded.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### Ovis2-5-2B

Layout PLCC: 0.812. Result ID: 69988ccb5698665ec4ee5de7.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7aa61e209131cea84620.md>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### Qwen3-VL-2B

Layout PLCC: 0.808. Result ID: 69988ccb5698665ec4ee5deb.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### mPLUG-Owl3-7B

Layout PLCC: 0.784. Result ID: 69988ccb5698665ec4ee5de8.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa746c1e209131cea61a16.md>)

[Implementation](<https://github.com/X-PLUG/mPLUG-Owl>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### InternVL3-5-8B

Layout PLCC: 0.774. Result ID: 69988ccb5698665ec4ee5dea.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"MLLMs (SFT-based)"}. This record does not assert independent reproduction.

### Q-Eval-Score

Layout PLCC: 0.265. Result ID: 69988ccb5698665ec4ee5de1.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa2.md>)

[Implementation](<https://github.com/zzc-1998/Q-Eval>)

Recorded evaluation configuration: {"Method Category":"Deep Learning-based NR-IQA"}. This record does not assert independent reproduction.

### gemini-2.5-pro

Layout PLCC: 0.246. Result ID: 69988ccb5698665ec4ee5de6.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"Commercial MLLMs"}. This record does not assert independent reproduction.

### DeQA-Score

Layout PLCC: 0.201. Result ID: 69988ccb5698665ec4ee5ddd.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77991e209131cea74442.md>)

Recorded evaluation configuration: {"Method Category":"Deep Learning-based NR-IQA"}. This record does not assert independent reproduction.

### CLIPIQA

Layout PLCC: 0.197. Result ID: 69988ccb5698665ec4ee5dd8.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dfd1e209131cea3d39b.md>)

[Implementation](<https://github.com/IceClear/CLIP-IQA>)

Recorded evaluation configuration: {"Method Category":"Traditional NR-IQA"}. This record does not assert independent reproduction.

### MANIQA

Layout PLCC: 0.194. Result ID: 69988ccb5698665ec4ee5dd9.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c551e209131cea3356a.md>)

[Implementation](<https://github.com/IIGROUP/MANIQA>)

Recorded evaluation configuration: {"Method Category":"Traditional NR-IQA"}. This record does not assert independent reproduction.

### gpt-5

Layout PLCC: 0.185. Result ID: 69988ccb5698665ec4ee5de3.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"Commercial MLLMs"}. This record does not assert independent reproduction.

### HyperIQA

Layout PLCC: 0.179. Result ID: 69988ccb5698665ec4ee5dda.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"Traditional NR-IQA"}. This record does not assert independent reproduction.

### DBCNN

Layout PLCC: 0.162. Result ID: 69988ccb5698665ec4ee5dd7.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65fe1e209131cea0dcc9.md>)

[Implementation](<https://github.com/zwx8981/BIQA_project>)

Recorded evaluation configuration: {"Method Category":"Traditional NR-IQA"}. This record does not assert independent reproduction.

### Compare2Score

Layout PLCC: 0.16. Result ID: 69988ccb5698665ec4ee5ddc.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa736b1e209131cea5c604.md>)

Recorded evaluation configuration: {"Method Category":"Deep Learning-based NR-IQA"}. This record does not assert independent reproduction.

### Q-SiT

Layout PLCC: 0.141. Result ID: 69988ccb5698665ec4ee5dde.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77671e209131cea73113.md>)

Recorded evaluation configuration: {"Method Category":"Deep Learning-based NR-IQA"}. This record does not assert independent reproduction.

### MUSIQ

Layout PLCC: 0.121. Result ID: 69988ccb5698665ec4ee5dd6.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a921e209131cea28fa7.md>)

[Implementation](<https://github.com/google-research/google-research/tree/master/musiq>)

Recorded evaluation configuration: {"Method Category":"Traditional NR-IQA"}. This record does not assert independent reproduction.

### qwen-vl-max

Layout PLCC: 0.088. Result ID: 69988ccb5698665ec4ee5de5.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"Commercial MLLMs"}. This record does not assert independent reproduction.

### Q-Align-aesthetics

Layout PLCC: 0.075. Result ID: 69988ccb5698665ec4ee5de0.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71a31e209131cea523a2.md>)

[Implementation](<https://github.com/Q-Future/Q-Align>)

Recorded evaluation configuration: {"Method Category":"Deep Learning-based NR-IQA"}. This record does not assert independent reproduction.

### Q-Align-quality

Layout PLCC: 0.057. Result ID: 69988ccb5698665ec4ee5ddf.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71a31e209131cea523a2.md>)

[Implementation](<https://github.com/Q-Future/Q-Align>)

Recorded evaluation configuration: {"Method Category":"Deep Learning-based NR-IQA"}. This record does not assert independent reproduction.

### Q-Insight

Layout PLCC: 0.053. Result ID: 69988ccb5698665ec4ee5ddb.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78b51e209131cea79e58.md>)

Recorded evaluation configuration: {"Method Category":"Deep Learning-based NR-IQA"}. This record does not assert independent reproduction.

### gpt-4o

Layout PLCC: 0.017. Result ID: 69988ccb5698665ec4ee5de2.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"Commercial MLLMs"}. This record does not assert independent reproduction.

### claude3.5-sonnet

Layout PLCC: -0.001. Result ID: 69988ccb5698665ec4ee5de4.

[Reporting paper](<https://www.sota2.com/research/paper/6932675cd6b0dd3342934591.md>)

Recorded evaluation configuration: {"Method Category":"Commercial MLLMs"}. This record does not assert independent reproduction.

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