# Visual Question Answering on VizWiz 2022 (test-dev)

Source: [Visual Question Answering on VizWiz 2022 (test-dev)](<https://www.sota2.com/research/sota/visual-question-answering-on-vizwiz-2022-test-dev>)

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

Data updated: 2026-02-26T05:28:20.978Z

Evaluation of Visual Question Answering performance on the test-development split of the VizWiz 2022 dataset, typically measured by Accuracy.

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 Question Answering |
| Dataset (including any stated split) | VizWiz 2022 (test-dev) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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-question-answering-on-vizwiz-2022-test-dev.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Ensemble | 61.64 | Not recorded | 68fa6cc21e209131cea35d8f |
| RN50x64 | 60.73 | Not recorded | 68fa6cc21e209131cea35d8f |
| ViT-L/14@336px | 60.66 | Not recorded | 68fa6cc21e209131cea35d8f |

### Ensemble

Accuracy: 61.64. Result ID: 698f3ca53a6c314849f7642b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cc21e209131cea35d8f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cc21e209131cea35d8f.md>)

Recorded evaluation configuration: {"Components":"RN50x64 + ViT-L/14@336px"}. This record does not assert independent reproduction.

### RN50x64

Accuracy: 60.73. Result ID: 698f3ca53a6c314849f76429.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cc21e209131cea35d8f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cc21e209131cea35d8f.md>)

Recorded evaluation configuration: {"Backbone":"RN50x64 (CLIP)","Resolution":"448x448","Frozen Backbone":"true"}. This record does not assert independent reproduction.

### ViT-L/14@336px

Accuracy: 60.66. Result ID: 698f3ca53a6c314849f7642a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6cc21e209131cea35d8f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cc21e209131cea35d8f.md>)

Recorded evaluation configuration: {"Backbone":"ViT-L/14 (CLIP)","Resolution":"336x336","Frozen Backbone":"true"}. This record does not assert independent reproduction.

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