Visual QA on EndoVis18 VQA standard (test)
54.8BLEU-4Object Tokens (Ours)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Object Tokens (Ours)2026.06 | 54.8 | 6.049 | 81.57 | |
| EndoChatIn-domain model=true2026.06 | 52.2 | 5.99 | 79.62 | |
| Gemma-3-12BFine-tuned=true2026.06 | 52.15 | 5.919 | 78.36 | |
| LLaMA-3.2-11BFine-tuned=true2026.06 | 49.87 | 4.549 | 78.36 | |
| GLM-4V-9BFine-tuned=true2026.06 | 37.42 | 3.16 | 54.47 |