# Image-caption retrieval on ROCO (Perturbation Analysis)

Source: [Image-caption retrieval on ROCO (Perturbation Analysis)](<https://www.sota2.com/research/sota/image-caption-retrieval-on-roco-perturbation-analysis>)

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

Data updated: 2026-02-26T05:59:40.253Z

Evaluation of Image-caption retrieval performance on the ROCO dataset, specifically measuring standard R@10, perturbed R@10, and the resulting change in mean rank.

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 | Image-caption retrieval |
| Dataset (including any stated split) | ROCO |
| Metric | R@10 (Std) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [R@10 (Std)](<https://www.sota2.com/research/sota/image-caption-retrieval-on-roco-perturbation-analysis.md?metric=R%4010%20(Std)>): Higher is better; unit: PERCENT.

- [R@10 (Pert.)](<https://www.sota2.com/research/sota/image-caption-retrieval-on-roco-perturbation-analysis.md?metric=R%4010%20(Pert.)>): Higher is better; unit: PERCENT.

- [Δ Mean Rank](<https://www.sota2.com/research/sota/image-caption-retrieval-on-roco-perturbation-analysis.md?metric=%CE%94%20Mean%20Rank>): Lower is better; unit: SCALAR.

## Recorded results

| Method | R@10 (Std) | Result date | Report paper |
| --- | --- | --- | --- |
| Robust-MMR | 66.1 | Not recorded | 699bdf1909c33900fcd804ee |
| Robust-MMR | 61.2 | Not recorded | 699bdf1909c33900fcd804ee |
| METER | 45.1 | Not recorded | 699bdf1909c33900fcd804ee |
| ViLT | 43.2 | Not recorded | 699bdf1909c33900fcd804ee |

### Robust-MMR

R@10 (Std): 66.1. Result ID: 699cde6b5698665ec47f22a7.

[Reporting paper](<https://www.sota2.com/research/paper/699bdf1909c33900fcd804ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/699bdf1909c33900fcd804ee.md>)

Recorded evaluation configuration: {"evaluation protocol":"fine-tuned"}. This record does not assert independent reproduction.

### Robust-MMR

R@10 (Std): 61.2. Result ID: 699cde6b5698665ec47f22a6.

[Reporting paper](<https://www.sota2.com/research/paper/699bdf1909c33900fcd804ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/699bdf1909c33900fcd804ee.md>)

Recorded evaluation configuration: {"evaluation protocol":"zero-shot"}. This record does not assert independent reproduction.

### METER

R@10 (Std): 45.1. Result ID: 699cde6b5698665ec47f22a5.

[Reporting paper](<https://www.sota2.com/research/paper/699bdf1909c33900fcd804ee.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### ViLT

R@10 (Std): 43.2. Result ID: 699cde6b5698665ec47f22a4.

[Reporting paper](<https://www.sota2.com/research/paper/699bdf1909c33900fcd804ee.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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