# Text-to-Image Retrieval on Retrieval VQA Visual Genome (test)

Source: [Text-to-Image Retrieval on Retrieval VQA Visual Genome (test)](<https://www.sota2.com/research/sota/text-to-image-retrieval-on-retrieval-vqa-visual-genome-test>)

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

Data updated: 2026-03-27T14:08:49.948Z

Evaluation of Text-to-Image Retrieval performance on the test split of the Retrieval VQA Visual Genome dataset, measured by accuracy across specific visual attributes (Count, Color, Attribute, Shape, Relation).

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 | Text-to-Image Retrieval |
| Dataset (including any stated split) | Retrieval VQA Visual Genome (test) |
| Metric | Count Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

- [Count Accuracy](<https://www.sota2.com/research/sota/text-to-image-retrieval-on-retrieval-vqa-visual-genome-test.md?metric=Count%20Accuracy>): Higher is better; unit: PERCENT.

- [Color Accuracy](<https://www.sota2.com/research/sota/text-to-image-retrieval-on-retrieval-vqa-visual-genome-test.md?metric=Color%20Accuracy>): Higher is better; unit: PERCENT.

- [Attribute Accuracy](<https://www.sota2.com/research/sota/text-to-image-retrieval-on-retrieval-vqa-visual-genome-test.md?metric=Attribute%20Accuracy>): Higher is better; unit: PERCENT.

- [Shape Accuracy](<https://www.sota2.com/research/sota/text-to-image-retrieval-on-retrieval-vqa-visual-genome-test.md?metric=Shape%20Accuracy>): Higher is better; unit: PERCENT.

- [Relation Accuracy](<https://www.sota2.com/research/sota/text-to-image-retrieval-on-retrieval-vqa-visual-genome-test.md?metric=Relation%20Accuracy>): Higher is better; unit: PERCENT.

- [Average Accuracy](<https://www.sota2.com/research/sota/text-to-image-retrieval-on-retrieval-vqa-visual-genome-test.md?metric=Average%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Count Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| GoldiCLIP | 66.8 | Not recorded | 69c60ee30636e5bf4afc8367 |
| SigLIP 2 | 66.6 | Not recorded | 69c60ee30636e5bf4afc8367 |
| PE\_core | 65.1 | Not recorded | 69c60ee30636e5bf4afc8367 |
| SigLIP | 64.1 | Not recorded | 69c60ee30636e5bf4afc8367 |
| OpenCLIP | 62.1 | Not recorded | 69c60ee30636e5bf4afc8367 |
| COSMOS | 60.7 | Not recorded | 69c60ee30636e5bf4afc8367 |
| FLAIR | 57.6 | Not recorded | 69c60ee30636e5bf4afc8367 |

### GoldiCLIP

Count Accuracy: 66.8. Result ID: 69c68e2147f34ce924b0ce74.

[Reporting paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

Recorded evaluation configuration: {"Data size":"30M","Zero-shot":"true"}. This record does not assert independent reproduction.

### SigLIP 2

Count Accuracy: 66.6. Result ID: 69c68e2147f34ce924b0ce78.

[Reporting paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa770a1e209131cea70ad4.md>)

Recorded evaluation configuration: {"Data size":"10B","Zero-shot":"true"}. This record does not assert independent reproduction.

### PE\_core

Count Accuracy: 65.1. Result ID: 69c68e2147f34ce924b0ce76.

[Reporting paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78411e209131cea77734.md>)

Recorded evaluation configuration: {"Data size":"5B","Zero-shot":"true"}. This record does not assert independent reproduction.

### SigLIP

Count Accuracy: 64.1. Result ID: 69c68e2147f34ce924b0ce77.

[Reporting paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70ae1e209131cea4c7a4.md>)

Recorded evaluation configuration: {"Data size":"10B","Zero-shot":"true"}. This record does not assert independent reproduction.

### OpenCLIP

Count Accuracy: 62.1. Result ID: 69c68e2147f34ce924b0ce75.

[Reporting paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74261e209131cea6003e.md>)

[Implementation](<https://github.com/LAION-AI/scaling-laws-openclip>)

Recorded evaluation configuration: {"Data size":"2B","Zero-shot":"true"}. This record does not assert independent reproduction.

### COSMOS

Count Accuracy: 60.7. Result ID: 69c68e2147f34ce924b0ce73.

[Reporting paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77a01e209131cea74770.md>)

[Implementation](<https://www.sota2.com/ExplainableML/cosmos>)

Recorded evaluation configuration: {"Data size":"30M","Zero-shot":"true"}. This record does not assert independent reproduction.

### FLAIR

Count Accuracy: 57.6. Result ID: 69c68e2147f34ce924b0ce72.

[Reporting paper](<https://www.sota2.com/research/paper/69c60ee30636e5bf4afc8367.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76001e209131cea6ad67.md>)

[Implementation](<https://github.com/ExplainableML/flair>)

Recorded evaluation configuration: {"Data size":"30M","Zero-shot":"true"}. This record does not assert independent reproduction.

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