# Weakly Supervised Grounding on Visual Genome (VG) (test)

Source: [Weakly Supervised Grounding on Visual Genome (VG) (test)](<https://www.sota2.com/research/sota/weakly-supervised-grounding-on-visual-genome-vg-test>)

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

Data updated: 2026-02-26T05:34:23.765Z

Evaluation of Weakly Supervised Grounding performance on the test split of the Visual Genome dataset, measured using Accuracy (Pointing Game).

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 | Weakly Supervised Grounding |
| Dataset (including any stated split) | Visual Genome (VG) (test) |
| Metric | Accuracy (Pointing Game) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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 (Pointing Game)](<https://www.sota2.com/research/sota/weakly-supervised-grounding-on-visual-genome-vg-test.md?metric=Accuracy%20(Pointing%20Game)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy (Pointing Game) | Result date | Report paper |
| --- | --- | --- | --- |
| GbS | 55.91 | Not recorded | 68fa69c81e209131cea23f3b |
| MG | 55.16 | Not recorded | 68fa69c81e209131cea23f3b |
| GbS ensemble | 54.55 | Not recorded | 68fa69c81e209131cea23f3b |
| GbS | 53.4 | Not recorded | 68fa69c81e209131cea23f3b |
| GbS | 52.7 | Not recorded | 68fa69c81e209131cea23f3b |
| MG | 52.33 | Not recorded | 68fa69c81e209131cea23f3b |
| GbS | 52 | Not recorded | 68fa69c81e209131cea23f3b |
| MG | 48.76 | Not recorded | 68fa69c81e209131cea23f3b |
| MG | 47.94 | Not recorded | 68fa69c81e209131cea23f3b |
| SSS | 30.03 | Not recorded | 68fa69c81e209131cea23f3b |
| VGLS | 24.4 | Not recorded | 68fa69c81e209131cea23f3b |
| Baseline (Center) | 20.55 | Not recorded | 68fa69c81e209131cea23f3b |
| TD | 19.31 | Not recorded | 68fa69c81e209131cea23f3b |
| FCVC | 14.03 | Not recorded | 68fa69c81e209131cea23f3b |
| Baseline (Random) | 11.15 | Not recorded | 68fa69c81e209131cea23f3b |

### GbS

Accuracy (Pointing Game): 55.91. Result ID: 698de154ffc3901de099540e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

Recorded evaluation configuration: {"Backbone":"PNASNet","Training":"VG"}. This record does not assert independent reproduction.

### MG

Accuracy (Pointing Game): 55.16. Result ID: 698de154ffc3901de099540d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65db1e209131cea0cd44.md>)

[Implementation](<https://github.com/hassanhub/MultiGrounding>)

Recorded evaluation configuration: {"Backbone":"PNASNet","Training":"VG"}. This record does not assert independent reproduction.

### GbS ensemble

Accuracy (Pointing Game): 54.55. Result ID: 698de154ffc3901de0995415.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

Recorded evaluation configuration: {"Backbone":"Ensemble","Training":"MS-COCO"}. This record does not assert independent reproduction.

### GbS

Accuracy (Pointing Game): 53.4. Result ID: 698de154ffc3901de099540c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

Recorded evaluation configuration: {"Backbone":"VGG","Training":"VG"}. This record does not assert independent reproduction.

### GbS

Accuracy (Pointing Game): 52.7. Result ID: 698de154ffc3901de0995414.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

Recorded evaluation configuration: {"Backbone":"PNASNet","Training":"MS-COCO"}. This record does not assert independent reproduction.

### MG

Accuracy (Pointing Game): 52.33. Result ID: 698de154ffc3901de0995413.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65db1e209131cea0cd44.md>)

[Implementation](<https://github.com/hassanhub/MultiGrounding>)

Recorded evaluation configuration: {"Backbone":"PNASNet","Training":"MS-COCO"}. This record does not assert independent reproduction.

### GbS

Accuracy (Pointing Game): 52. Result ID: 698de154ffc3901de0995412.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

Recorded evaluation configuration: {"Backbone":"VGG","Training":"MS-COCO"}. This record does not assert independent reproduction.

### MG

Accuracy (Pointing Game): 48.76. Result ID: 698de154ffc3901de099540b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65db1e209131cea0cd44.md>)

[Implementation](<https://github.com/hassanhub/MultiGrounding>)

Recorded evaluation configuration: {"Backbone":"VGG","Training":"VG"}. This record does not assert independent reproduction.

### MG

Accuracy (Pointing Game): 47.94. Result ID: 698de154ffc3901de0995411.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65db1e209131cea0cd44.md>)

[Implementation](<https://github.com/hassanhub/MultiGrounding>)

Recorded evaluation configuration: {"Backbone":"VGG","Training":"MS-COCO"}. This record does not assert independent reproduction.

### SSS

Accuracy (Pointing Game): 30.03. Result ID: 698de154ffc3901de099540a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65121e209131cea07f95.md>)

Recorded evaluation configuration: {"Backbone":"VGG","Training":"VG"}. This record does not assert independent reproduction.

### VGLS

Accuracy (Pointing Game): 24.4. Result ID: 698de154ffc3901de0995410.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa634d1e209131ce9fde19.md>)

Recorded evaluation configuration: {"Backbone":"VGG","Training":"MS-COCO"}. This record does not assert independent reproduction.

### Baseline (Center)

Accuracy (Pointing Game): 20.55. Result ID: 698de154ffc3901de0995408.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

Recorded evaluation configuration: {"Backbone":"Center","Training":"N/A"}. This record does not assert independent reproduction.

### TD

Accuracy (Pointing Game): 19.31. Result ID: 698de154ffc3901de0995409.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62a51e209131ce9fa881.md>)

Recorded evaluation configuration: {"Backbone":"Inception-2","Training":"VG"}. This record does not assert independent reproduction.

### FCVC

Accuracy (Pointing Game): 14.03. Result ID: 698de154ffc3901de099540f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa624c1e209131ce9f8b1e.md>)

[Implementation](<http://research.microsoft.com/image_captioning>)

Recorded evaluation configuration: {"Backbone":"VGG","Training":"MS-COCO"}. This record does not assert independent reproduction.

### Baseline (Random)

Accuracy (Pointing Game): 11.15. Result ID: 698de154ffc3901de0995407.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69c81e209131cea23f3b.md>)

Recorded evaluation configuration: {"Backbone":"Random","Training":"N/A"}. This record does not assert independent reproduction.

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