# Visual Grounding on RefCOCO UMD (val)

Source: [Visual Grounding on RefCOCO UMD (val)](<https://www.sota2.com/research/sota/visual-grounding-on-refcoco-umd-val>)

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

Data updated: 2026-02-26T05:31:57.719Z

Evaluation of Visual Grounding performance on the validation split of the RefCOCO UMD dataset.

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 | Visual Grounding |
| Dataset (including any stated split) | RefCOCO UMD (val) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 12 of 12 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/visual-grounding-on-refcoco-umd-val.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Base Model | 93 | Not recorded | 6937acf86cfd488761a4e6ac |
| ECVGPO | 92.15 | Not recorded | 6937acf86cfd488761a4e6ac |
| GRPO | 91.65 | Not recorded | 6937acf86cfd488761a4e6ac |
| ECVGPO | 90.55 | Not recorded | 6937acf86cfd488761a4e6ac |
| GRPO | 90.16 | Not recorded | 6937acf86cfd488761a4e6ac |
| Base Model | 89.8 | Not recorded | 6937acf86cfd488761a4e6ac |
| ECVGPO | 89.49 | Not recorded | 6937acf86cfd488761a4e6ac |
| GRPO | 89.21 | Not recorded | 6937acf86cfd488761a4e6ac |
| SFT | 88.75 | Not recorded | 6937acf86cfd488761a4e6ac |
| SFT | 88.3 | Not recorded | 6937acf86cfd488761a4e6ac |
| Base Model | 88.05 | Not recorded | 6937acf86cfd488761a4e6ac |
| SFT | 86.65 | Not recorded | 6937acf86cfd488761a4e6ac |

### Base Model

Accuracy: 93. Result ID: 699880765698665ec4e1a147.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct"}. This record does not assert independent reproduction.

### ECVGPO

Accuracy: 92.15. Result ID: 699880765698665ec4e1a14a.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct"}. This record does not assert independent reproduction.

### GRPO

Accuracy: 91.65. Result ID: 699880765698665ec4e1a149.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct"}. This record does not assert independent reproduction.

### ECVGPO

Accuracy: 90.55. Result ID: 699880765698665ec4e1a142.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B-Instruct"}. This record does not assert independent reproduction.

### GRPO

Accuracy: 90.16. Result ID: 699880765698665ec4e1a141.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B-Instruct"}. This record does not assert independent reproduction.

### Base Model

Accuracy: 89.8. Result ID: 699880765698665ec4e1a143.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"InternVL3-2B-Instruct"}. This record does not assert independent reproduction.

### ECVGPO

Accuracy: 89.49. Result ID: 699880765698665ec4e1a146.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"InternVL3-2B-Instruct"}. This record does not assert independent reproduction.

### GRPO

Accuracy: 89.21. Result ID: 699880765698665ec4e1a145.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Backbone":"InternVL3-2B-Instruct"}. This record does not assert independent reproduction.

### SFT

Accuracy: 88.75. Result ID: 699880765698665ec4e1a140.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B-Instruct"}. This record does not assert independent reproduction.

### SFT

Accuracy: 88.3. Result ID: 699880765698665ec4e1a148.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"InternVL3-8B-Instruct"}. This record does not assert independent reproduction.

### Base Model

Accuracy: 88.05. Result ID: 699880765698665ec4e1a13f.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B-Instruct"}. This record does not assert independent reproduction.

### SFT

Accuracy: 86.65. Result ID: 699880765698665ec4e1a144.

[Reporting paper](<https://www.sota2.com/research/paper/6937acf86cfd488761a4e6ac.md>)

Recorded evaluation configuration: {"Backbone":"InternVL3-2B-Instruct"}. This record does not assert independent reproduction.

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