# Region Captioning on Visual Genome (val)

Source: [Region Captioning on Visual Genome (val)](<https://www.sota2.com/research/sota/region-captioning-on-visual-genome-val>)

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

Data updated: 2026-02-26T05:49:48.612Z

Evaluation of Region Captioning performance on the validation split of the Visual Genome dataset, measured using standard captioning metrics like BLEU, METEOR, ROUGE, and CIDEr.

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 | Region Captioning |
| Dataset (including any stated split) | Visual Genome (val) |
| Metric | METEOR |
| 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

- [METEOR](<https://www.sota2.com/research/sota/region-captioning-on-visual-genome-val.md?metric=METEOR>): Higher is better; unit: SCALAR.

- [BLEU@4](<https://www.sota2.com/research/sota/region-captioning-on-visual-genome-val.md?metric=BLEU%404>): Higher is better; unit: SCALAR.

- [ROUGE](<https://www.sota2.com/research/sota/region-captioning-on-visual-genome-val.md?metric=ROUGE>): Higher is better; unit: SCALAR.

- [CIDEr](<https://www.sota2.com/research/sota/region-captioning-on-visual-genome-val.md?metric=CIDEr>): Higher is better; unit: SCALAR.

## Recorded results

| Method | METEOR | Result date | Report paper |
| --- | --- | --- | --- |
| GPT4RoI-13B | 17.6 | Not recorded | 68fa794e1e209131cea7d037 |
| GPT4RoI-7B | 17.4 | Not recorded | 68fa794e1e209131cea7d037 |
| GRIT | 17.1 | Not recorded | 68fa794e1e209131cea7d037 |
| GPT4RoI-7B | 16.5 | Not recorded | 68fa794e1e209131cea7d037 |
| Shikra-7B | 15.2 | Not recorded | 68fa794e1e209131cea7d037 |

### GPT4RoI-13B

METEOR: 17.6. Result ID: 6995d9003badfaffe6df32a1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Implementation](<https://github.com/jshilong/GPT4RoI>)

Recorded evaluation configuration: {"Input":"Ground truth bounding boxes","fine-tuned\_on\_VG":true}. This record does not assert independent reproduction.

### GPT4RoI-7B

METEOR: 17.4. Result ID: 6995d9003badfaffe6df32a0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Implementation](<https://github.com/jshilong/GPT4RoI>)

Recorded evaluation configuration: {"Input":"Ground truth bounding boxes","fine-tuned\_on\_VG":true}. This record does not assert independent reproduction.

### GRIT

METEOR: 17.1. Result ID: 6995d9003badfaffe6df329d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e231e209131cea3e161.md>)

[Implementation](<https://github.com/JialianW/GRiT>)

Recorded evaluation configuration: {"Input":"Ground truth bounding boxes"}. This record does not assert independent reproduction.

### GPT4RoI-7B

METEOR: 16.5. Result ID: 6995d9003badfaffe6df329f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Implementation](<https://github.com/jshilong/GPT4RoI>)

Recorded evaluation configuration: {"Input":"Ground truth bounding boxes","fine-tuned\_on\_VG":false}. This record does not assert independent reproduction.

### Shikra-7B

METEOR: 15.2. Result ID: 6995d9003badfaffe6df329e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa794e1e209131cea7d037.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fe61e209131cea47fbc.md>)

[Implementation](<https://github.com/shikras/shikra>)

Recorded evaluation configuration: {"Input":"Ground truth bounding boxes","concurrent\_work":true}. This record does not assert independent reproduction.

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