# Knowledge-aware Visual Question Answering on KVQA Paraphrased PRP questions Oracle setting (test)

Source: [Knowledge-aware Visual Question Answering on KVQA Paraphrased PRP questions Oracle setting (test)](<https://www.sota2.com/research/sota/knowledge-aware-visual-question-answering-on-kvqa-paraphrased-prp-questions-oracle-setting-test>)

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

Data updated: 2026-02-26T05:51:49.973Z

Evaluation of Knowledge-aware Visual Question Answering performance on the test split of the KVQA Paraphrased PRP dataset, specifically under the Oracle setting, reporting hop-based accuracies.

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-aware Visual Question Answering |
| Dataset (including any stated split) | KVQA Paraphrased PRP questions Oracle setting (test) |
| Metric | Mean Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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

- [Mean Accuracy](<https://www.sota2.com/research/sota/knowledge-aware-visual-question-answering-on-kvqa-paraphrased-prp-questions-oracle-setting-test.md?metric=Mean%20Accuracy>): Higher is better; unit: PERCENT.

- [1-hop Accuracy](<https://www.sota2.com/research/sota/knowledge-aware-visual-question-answering-on-kvqa-paraphrased-prp-questions-oracle-setting-test.md?metric=1-hop%20Accuracy>): Higher is better; unit: PERCENT.

- [2-hop Accuracy](<https://www.sota2.com/research/sota/knowledge-aware-visual-question-answering-on-kvqa-paraphrased-prp-questions-oracle-setting-test.md?metric=2-hop%20Accuracy>): Higher is better; unit: PERCENT.

- [3-hop Accuracy](<https://www.sota2.com/research/sota/knowledge-aware-visual-question-answering-on-kvqa-paraphrased-prp-questions-oracle-setting-test.md?metric=3-hop%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Mean Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Hypergraph Transformer | 89.7 | Not recorded | 68fa6c571e209131cea3367d |
| BAN | 83.8 | Not recorded | 68fa6c571e209131cea3367d |
| MemNN† | 77.3 | Not recorded | 68fa6c571e209131cea3367d |
| HAN | 77.3 | Not recorded | 68fa6c571e209131cea3367d |
| GGNN | 73.8 | Not recorded | 68fa6c571e209131cea3367d |
| GCN | 66.7 | Not recorded | 68fa6c571e209131cea3367d |
| MemNN | 59.2 | Not recorded | 68fa6c571e209131cea3367d |
| BLSTM | 51 | Not recorded | 68fa6c571e209131cea3367d |

### Hypergraph Transformer

Mean Accuracy: 89.7. Result ID: 698f05633a6c314849f4d898.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

[Implementation](<https://github.com/yujungheo/kbvqa-public>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak"}. This record does not assert independent reproduction.

### BAN

Mean Accuracy: 83.8. Result ID: 698f05633a6c314849f4d897.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64fd1e209131cea07a2c.md>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak"}. This record does not assert independent reproduction.

### MemNN†

Mean Accuracy: 77.3. Result ID: 698f05633a6c314849f4d895.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa622f1e209131ce9f80a6.md>)

[Implementation](<https://github.com/facebook/MemNN>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak","implementation":"re-implemented"}. This record does not assert independent reproduction.

### HAN

Mean Accuracy: 77.3. Result ID: 698f05633a6c314849f4d896.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak"}. This record does not assert independent reproduction.

### GGNN

Mean Accuracy: 73.8. Result ID: 698f05633a6c314849f4d894.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63a31e209131ce9ffef4.md>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak"}. This record does not assert independent reproduction.

### GCN

Mean Accuracy: 66.7. Result ID: 698f05633a6c314849f4d893.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa631d1e209131ce9fce45.md>)

[Implementation](<https://github.com/tkipf/gcn>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak"}. This record does not assert independent reproduction.

### MemNN

Mean Accuracy: 59.2. Result ID: 698f05633a6c314849f4d892.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa622f1e209131ce9f80a6.md>)

[Implementation](<https://github.com/facebook/MemNN>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak"}. This record does not assert independent reproduction.

### BLSTM

Mean Accuracy: 51. Result ID: 698f05633a6c314849f4d891.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c571e209131cea3367d.md>)

Recorded evaluation configuration: {"Oracle setting":true,"Supervision":"weak"}. This record does not assert independent reproduction.

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