Knowledge-based Visual Question Answering on A-OKVQA (test)
59.5AccuracyBLIP VQAv2
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| BLIP VQAv2Fine-tuned on=VQAv22023.06 | 59.5 | — | — | |
| BLIP VQAv2 + SelTDAFine-tuned on=VQAv2, Augmentation=SelTDA2023.06 | 59.5 | 0 | 5.8 | |
| BLIP + SelTDAAugmentation=SelTDA2023.06 | 54.5 | — | 0.8 | |
| GPV-22023.06 | 53.7 | — | — | |
| KRISP2023.06 | 42.2 | — | — | |
| LXMERT2023.06 | 41.6 | — | — | |
| ViLBERT2023.06 | 41.5 | — | — |