Visual Question Answering on VQA-CE
39.24AccuracyVilBERT
Evaluation results
| Method | Links | |
|---|---|---|
| VilBERTpretrained=Conceptual Caption, fine-tuned=VQA v22021.04 | 39.24 | |
| RandImgBase architecture=UpDown [3]2021.04 | 34.41 | |
| LMH+CSSBase architecture=UpDown [3]2021.04 | 34.36 | |
| LfFBase architecture=UpDown [3]2021.04 | 34.27 | |
| LMHBase architecture=UpDown [3]2021.04 | 34.26 | |
| UpDown2021.04 | 33.91 | |
| ESRBase architecture=UpDown [3]2021.04 | 33.26 | |
| LMH + RMFEBase architecture=UpDown [3]2021.04 | 33.14 | |
| BLOCK2021.04 | 32.91 | |
| RUBiBase architecture=UpDown [3]2021.04 | 32.25 | |
| SANfeatures=grid2021.04 | 26.64 | |
| Question-Only2021.04 | 11.59 | |
| Image-Only2021.04 | 1.58 | |
| Shortcuts2021.04 | 0 |