EndoVis-VQA
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Binary safety hallucination detection | EndoVis18-VQA Out-of-template (val) | Accuracy98 | 50 | |
| Surgical Video Question Answering | EndoVis18-VQA Out-of-template | BLEU-428.15 | 20 | |
| Surgical Visual Question Answering | EndoVis18-VQA | FScore62.04 | 7 | |
| Medical Visual Question Answering | EndoVis-VQA | Accuracy54.2 | 6 | |
| Description | EndoVis18 VQA standard (test) | GPT-4 Score81.2 | 5 | |
| Region-Based QA | EndoVis18 VQA standard (test) | BLEU-460.73 | 5 | |
| Visual QA | EndoVis18 VQA standard (test) | BLEU-454.8 | 5 | |
| Single-Phrase Grounded QA | EndoVis18 VQA standard (test) | Accuracy76.67 | 5 | |
| Visual Question Answering | EndoVis18-VQA Out-of-template (val) | BLEU0.474 | 5 |
Showing 9 of 9 rows