# Knowledge-based Visual Question Answering on A-OKVQA (test)

Source: [Knowledge-based Visual Question Answering on A-OKVQA (test)](<https://www.sota2.com/research/sota/knowledge-based-visual-question-answering-on-a-okvqa-test>)

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

Data updated: 2026-02-26T05:28:58.274Z

Evaluation of model performance on the test split of the A-OKVQA dataset specifically targeting Knowledge-based Visual Question Answering capabilities.

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 | Knowledge-based Visual Question Answering |
| Dataset (including any stated split) | A-OKVQA (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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/knowledge-based-visual-question-answering-on-a-okvqa-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Gain vs Baseline (%)](<https://www.sota2.com/research/sota/knowledge-based-visual-question-answering-on-a-okvqa-test.md?metric=Gain%20vs%20Baseline%20(%25)>): Higher is better; unit: PERCENT.

- [Gain vs Best Prior Work (%)](<https://www.sota2.com/research/sota/knowledge-based-visual-question-answering-on-a-okvqa-test.md?metric=Gain%20vs%20Best%20Prior%20Work%20(%25)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| BLIP VQAv2 | 59.5 | Not recorded | 68fa6f9c1e209131cea4695d |
| BLIP VQAv2 + SelTDA | 59.5 | Not recorded | 68fa6f9c1e209131cea4695d |
| BLIP + SelTDA | 54.5 | Not recorded | 68fa6f9c1e209131cea4695d |
| GPV-2 | 53.7 | Not recorded | 68fa6f9c1e209131cea4695d |
| KRISP | 42.2 | Not recorded | 68fa6f9c1e209131cea4695d |
| LXMERT | 41.6 | Not recorded | 68fa6f9c1e209131cea4695d |
| ViLBERT | 41.5 | Not recorded | 68fa6f9c1e209131cea4695d |

### BLIP VQAv2

Accuracy: 59.5. Result ID: 6990e3bf3a6c3148490410b9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bd81e209131cea30874.md>)

[Implementation](<https://github.com/salesforce/BLIP>)

Recorded evaluation configuration: {"Fine-tuned on":"VQAv2"}. This record does not assert independent reproduction.

### BLIP VQAv2 + SelTDA

Accuracy: 59.5. Result ID: 6990e3bf3a6c3148490410bb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Implementation](<https://github.com/codezakh/SelTDA>)

Recorded evaluation configuration: {"Fine-tuned on":"VQAv2","Augmentation":"SelTDA"}. This record does not assert independent reproduction.

### BLIP + SelTDA

Accuracy: 54.5. Result ID: 6990e3bf3a6c3148490410ba.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Implementation](<https://github.com/codezakh/SelTDA>)

Recorded evaluation configuration: {"Augmentation":"SelTDA"}. This record does not assert independent reproduction.

### GPV-2

Accuracy: 53.7. Result ID: 6990e3bf3a6c3148490410b8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d1a1e209131cea37c03.md>)

[Implementation](<https://prior.allenai.org/projects/gpv2>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### KRISP

Accuracy: 42.2. Result ID: 6990e3bf3a6c3148490410b7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68f21e209131cea1f307.md>)

[Implementation](<https://github.com/facebookresearch/mmf>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### LXMERT

Accuracy: 41.6. Result ID: 6990e3bf3a6c3148490410b6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66cc1e209131cea12644.md>)

[Implementation](<https://github.com/airsplay/lxmert>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### ViLBERT

Accuracy: 41.5. Result ID: 6990e3bf3a6c3148490410b5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f9c1e209131cea4695d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa661c1e209131cea0e5d5.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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