# Image Description on Image description

Source: [Image Description on Image description](<https://www.sota2.com/research/sota/image-description-on-image-description>)

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

Data updated: 2026-03-26T15:24:13.671Z

Evaluation of model performance on generating natural language descriptions for images, assessed using metrics such as BLEU, ROUGE-L, BERTScore, and SBERT Similarity.

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 Description |
| Dataset (including any stated split) | Image description |
| Metric | BLEU-1 |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [BLEU-1](<https://www.sota2.com/research/sota/image-description-on-image-description.md?metric=BLEU-1>): Higher is better; unit: SCALAR.

- [ROUGE-L F1](<https://www.sota2.com/research/sota/image-description-on-image-description.md?metric=ROUGE-L%20F1>): Higher is better; unit: SCALAR.

- [BERTScore F1](<https://www.sota2.com/research/sota/image-description-on-image-description.md?metric=BERTScore%20F1>): Higher is better; unit: SCALAR.

- [SBERT Similarity](<https://www.sota2.com/research/sota/image-description-on-image-description.md?metric=SBERT%20Similarity>): Higher is better; unit: SCALAR.

## Recorded results

| Method | BLEU-1 | Result date | Report paper |
| --- | --- | --- | --- |
| VOLMO-2B | 0.1741 | Not recorded | 69c4bdad0636e5bf4afc76b5 |
| LLaVA-Med | 0.1181 | Not recorded | 69c4bdad0636e5bf4afc76b5 |
| MedGemma-4B | 0.1083 | Not recorded | 69c4bdad0636e5bf4afc76b5 |
| MedGemma-27B | 0.0912 | Not recorded | 69c4bdad0636e5bf4afc76b5 |
| InternVL-2B | 0.0796 | Not recorded | 69c4bdad0636e5bf4afc76b5 |

### VOLMO-2B

BLEU-1: 0.1741. Result ID: 69c54f9247f34ce924af873f.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bdad0636e5bf4afc76b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c4bdad0636e5bf4afc76b5.md>)

Recorded evaluation configuration: {"Model":"VOLMO-2B"}. This record does not assert independent reproduction.

### LLaVA-Med

BLEU-1: 0.1181. Result ID: 69c54f9247f34ce924af873c.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bdad0636e5bf4afc76b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f8e1e209131cea46356.md>)

[Implementation](<https://aka.ms/llava-med>)

Recorded evaluation configuration: {"Model":"LLaVA-Med"}. This record does not assert independent reproduction.

### MedGemma-4B

BLEU-1: 0.1083. Result ID: 69c54f9247f34ce924af873d.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bdad0636e5bf4afc76b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79ff1e209131cea808a4.md>)

Recorded evaluation configuration: {"Model":"MedGemma-4B"}. This record does not assert independent reproduction.

### MedGemma-27B

BLEU-1: 0.0912. Result ID: 69c54f9247f34ce924af873e.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bdad0636e5bf4afc76b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79ff1e209131cea808a4.md>)

Recorded evaluation configuration: {"Model":"MedGemma-27B"}. This record does not assert independent reproduction.

### InternVL-2B

BLEU-1: 0.0796. Result ID: 69c54f9247f34ce924af873b.

[Reporting paper](<https://www.sota2.com/research/paper/69c4bdad0636e5bf4afc76b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71de1e209131cea5370d.md>)

[Implementation](<https://github.com/OpenGVLab/InternVL>)

Recorded evaluation configuration: {"Model":"InternVL-2B"}. This record does not assert independent reproduction.

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