Explanatory Visual Question Answering On Gqa
Métriques
BLEU-4
CIDEr
GQA-test
GQA-val
Grounding
METEOR
ROUGE-L
SPICE
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | BLEU-4 | CIDEr | GQA-test | GQA-val | Grounding | METEOR | ROUGE-L | SPICE |
|---|---|---|---|---|---|---|---|---|
| variational-causal-inference-network-for | 58.65 | 519.23 | 60.61 | 81.80 | 77.33 | 41.57 | 81.45 | 54.63 |
| rex-reasoning-aware-and-grounded-explanation | 54.79 | 466.01 | 58.15 | 78.19 | 70.79 | 39.51 | 79.41 | 49.98 |
| faithful-multimodal-explanation-for-visual | 42.45 | 357.10 | 56.92 | 65.17 | 33.52 | 34.46 | 73.51 | 40.35 |
| vqa-e-explaining-elaborating-and-enhancing | 42.56 | 358.20 | 57.24 | 65.19 | 31.29 | 34.51 | 73.59 | 40.39 |
| rex-reasoning-aware-and-grounded-explanation | 54.59 | 464.20 | 57.77 | 66.16 | 67.95 | 39.22 | 78.56 | 46.80 |