Visual Question Answering on ArtVQA (test)
0.5686Overall AccuracyVIKING_VLM + SelTDA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VIKING_VLM + SelTDA2023.06 | 0.5686 | 83.86 | |
| VIKING_VLM2023.06 | 0.559 | 81.9 | |
| VIKING2023.06 | 0.555 | 78.74 | |
| BAN2023.06 | 0.224 | — | |
| BLIP + SelTDABackbone=ViT-B/162023.06 | 0.2168 | 83.86 | |
| BLIPBackbone=ViT-B/16, Pre-training pairs=129M2023.06 | 0.2136 | 81.71 |