# Image Captioning on MS COCO (Coco, Avg %)

Source: [Image Captioning on MS COCO (Coco, Avg %)](<https://www.sota2.com/research/sota/image-captioning-on-ms-coco-coco-avg-percent>)

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

Data updated: 2026-05-04T09:28:13.203Z

Evaluation of Image Captioning performance on the MS COCO dataset, reporting results specifically for the 'Coco' and 'Avg %' metrics.

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 Captioning |
| Dataset (including any stated split) | MS COCO |
| Metric | COCO Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [COCO Score](<https://www.sota2.com/research/sota/image-captioning-on-ms-coco-coco-avg-percent.md?metric=COCO%20Score>): Higher is better; unit: PERCENT.

- [Average Score (%)](<https://www.sota2.com/research/sota/image-captioning-on-ms-coco-coco-avg-percent.md?metric=Average%20Score%20(%25)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | COCO Score | Result date | Report paper |
| --- | --- | --- | --- |
| PiToMe | 38.9 | Not recorded | 69f8281cc0f56ed843088afc |
| LightKV | 38.9 | Not recorded | 69f8281cc0f56ed843088afc |
| ToFu | 38.3 | Not recorded | 69f8281cc0f56ed843088afc |
| FastV | 33.9 | Not recorded | 69f8281cc0f56ed843088afc |
| ToMe | 32.9 | Not recorded | 69f8281cc0f56ed843088afc |
| Vanilla | 31.9 | Not recorded | 69f8281cc0f56ed843088afc |

### PiToMe

COCO Score: 38.9. Result ID: 69f866a747f34ce924dc6f1b.

[Reporting paper](<https://www.sota2.com/research/paper/69f8281cc0f56ed843088afc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-VL-7B-Instruct","Vision token retention rate":"55%"}. This record does not assert independent reproduction.

### LightKV

COCO Score: 38.9. Result ID: 69f866a747f34ce924dc6f1d.

[Reporting paper](<https://www.sota2.com/research/paper/69f8281cc0f56ed843088afc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f8281cc0f56ed843088afc.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-VL-7B-Instruct","Vision token retention rate":"55%"}. This record does not assert independent reproduction.

### ToFu

COCO Score: 38.3. Result ID: 69f866a747f34ce924dc6f1c.

[Reporting paper](<https://www.sota2.com/research/paper/69f8281cc0f56ed843088afc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71561e209131cea509f8.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-VL-7B-Instruct","Vision token retention rate":"55%"}. This record does not assert independent reproduction.

### FastV

COCO Score: 33.9. Result ID: 69f866a747f34ce924dc6f19.

[Reporting paper](<https://www.sota2.com/research/paper/69f8281cc0f56ed843088afc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74ae1e209131cea632b7.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-VL-7B-Instruct","Vision token retention rate":"55%"}. This record does not assert independent reproduction.

### ToMe

COCO Score: 32.9. Result ID: 69f866a747f34ce924dc6f1a.

[Reporting paper](<https://www.sota2.com/research/paper/69f8281cc0f56ed843088afc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Model":"Qwen2.5-VL-7B-Instruct","Vision token retention rate":"55%"}. This record does not assert independent reproduction.

### Vanilla

COCO Score: 31.9. Result ID: 69f866a747f34ce924dc6f18.

[Reporting paper](<https://www.sota2.com/research/paper/69f8281cc0f56ed843088afc.md>)

Recorded evaluation configuration: {"Model":"Qwen2.5-VL-7B-Instruct","Vision token retention rate":"55%"}. This record does not assert independent reproduction.

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