# Visual Question Answering on HGI SemiosisArt

Source: [Visual Question Answering on HGI SemiosisArt](<https://www.sota2.com/research/sota/visual-question-answering-on-hgi-semiosisart>)

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

Data updated: 2026-04-13T07:49:41.264Z

Evaluation of model performance on the HGI SemiosisArt dataset for the Visual Question Answering task, reporting accuracy.

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 | Visual Question Answering |
| Dataset (including any stated split) | HGI SemiosisArt |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/visual-question-answering-on-hgi-semiosisart.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Human (Expert) | 93.2 | Not recorded | 69dc788d7d0172140ec78ad6 |
| SemJudge (Gemini-Flash) | 92.4 | Not recorded | 69dc788d7d0172140ec78ad6 |
| SemJudge (Qwen-35B-A3B) | 91 | Not recorded | 69dc788d7d0172140ec78ad6 |
| SemJudge (Qwen-9B) | 86.1 | Not recorded | 69dc788d7d0172140ec78ad6 |
| Human (Non-expert) | 81.5 | Not recorded | 69dc788d7d0172140ec78ad6 |
| ArtCoT | 80.4 | Not recorded | 69dc788d7d0172140ec78ad6 |
| ArtiMuse | 67.7 | Not recorded | 69dc788d7d0172140ec78ad6 |
| LMM4LMM | 44 | Not recorded | 69dc788d7d0172140ec78ad6 |
| GalleryGPT | 26.1 | Not recorded | 69dc788d7d0172140ec78ad6 |
| Random Guess | 25.2 | Not recorded | 69dc788d7d0172140ec78ad6 |

### Human (Expert)

Accuracy: 93.2. Result ID: 69dc9fdf47f34ce924c3598c.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

Recorded evaluation configuration: {"Group":"Human Reference","Backbone":"-"}. This record does not assert independent reproduction.

### SemJudge (Gemini-Flash)

Accuracy: 92.4. Result ID: 69dc9fdf47f34ce924c3598f.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Implementation](<https://github.com/songrise/SemJudge>)

Recorded evaluation configuration: {"Group":"Semiosis Quality","Backbone":"Zero-shot MLLM"}. This record does not assert independent reproduction.

### SemJudge (Qwen-35B-A3B)

Accuracy: 91. Result ID: 69dc9fdf47f34ce924c3598e.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Implementation](<https://github.com/songrise/SemJudge>)

Recorded evaluation configuration: {"Group":"Semiosis Quality","Backbone":"Zero-shot MLLM"}. This record does not assert independent reproduction.

### SemJudge (Qwen-9B)

Accuracy: 86.1. Result ID: 69dc9fdf47f34ce924c3598d.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Implementation](<https://github.com/songrise/SemJudge>)

Recorded evaluation configuration: {"Group":"Semiosis Quality","Backbone":"Zero-shot MLLM"}. This record does not assert independent reproduction.

### Human (Non-expert)

Accuracy: 81.5. Result ID: 69dc9fdf47f34ce924c3598b.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

Recorded evaluation configuration: {"Group":"Human Reference","Backbone":"-"}. This record does not assert independent reproduction.

### ArtCoT

Accuracy: 80.4. Result ID: 69dc9fdf47f34ce924c35988.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7af61e209131cea85fca.md>)

Recorded evaluation configuration: {"Group":"Formal Analysis","Backbone":"Zero-shot MLLM"}. This record does not assert independent reproduction.

### ArtiMuse

Accuracy: 67.7. Result ID: 69dc9fdf47f34ce924c35989.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a7d1e209131cea836d3.md>)

Recorded evaluation configuration: {"Group":"Art Interpretation","Backbone":"Tuned MLLM"}. This record does not assert independent reproduction.

### LMM4LMM

Accuracy: 44. Result ID: 69dc9fdf47f34ce924c35987.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77f61e209131cea763fb.md>)

Recorded evaluation configuration: {"Group":"Rationale&Scoring","Backbone":"Tuned MLLM"}. This record does not assert independent reproduction.

### GalleryGPT

Accuracy: 26.1. Result ID: 69dc9fdf47f34ce924c3598a.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa744f1e209131cea60f40.md>)

Recorded evaluation configuration: {"Group":"Art Interpretation","Backbone":"Tuned MLLM"}. This record does not assert independent reproduction.

### Random Guess

Accuracy: 25.2. Result ID: 69dc9fdf47f34ce924c35986.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad6.md>)

Recorded evaluation configuration: {"Group":"Conventional Scorers","Backbone":"-"}. This record does not assert independent reproduction.

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