Medical Visual Question Answering on Medical-Diff-VQA 1.0 (test)
0.747BLEU-1Attention-guided encoder-decoder
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Attention-guided encoder-decoderBackbone=Swin-base (microsoft/swin-base-patch4-window12-384-in22k), DINO backbone=RAD-DINO, Input Resolution=384x3842026.06 | 0.747 | 0.7 | 0.703 | 2.011 | |
| VED2026.06 | 0.716 | 0.389 | 0.67 | 2.119 | |
| RegioMix2026.06 | 0.705 | 0.381 | 0.651 | 1.804 | |
| PLURAL2026.06 | 0.704 | 0.381 | 0.653 | 1.832 | |
| EKAID2026.06 | 0.628 | 0.339 | 0.557 | 1.027 | |
| IDCPCL2026.06 | 0.614 | 0.303 | 0.582 | 0.703 | |
| MCCFormers2026.06 | 0.214 | 0.319 | 0.34 | 0 |