Visual Question Answering On Vizwiz 2018 1
Metriken
number
other
overall
unanswerable
yes/no
Ergebnisse
Leistungsergebnisse verschiedener Modelle zu diesem Benchmark
Vergleichstabelle
| Modellname | number | other | overall | unanswerable | yes/no |
|---|---|---|---|---|---|
| Modell 1 | 20.24 | 23.7 | 45.53 | 82.02 | 59.85 |
| Modell 2 | 17.86 | 31.5 | 51.61 | 85.28 | 68.12 |
| Modell 3 | 20.24 | 26.97 | 47.6 | 82.58 | 58.91 |
| Modell 4 | 21.67 | 33.98 | 51.71 | 80.59 | 72.04 |
| Modell 5 | 22.14 | 26.83 | 47.32 | 82.22 | 56.61 |
| Modell 6 | 26.9 | 37.6 | 54.93 | 83.59 | 71.18 |
| Modell 7 | 23.57 | 34.81 | 52.23 | 80.48 | 72.8 |
| decoupled-box-proposal-and-featurization-with | 28.81 | 35.41 | 53.68 | 84.03 | 68.12 |
| lxmert-learning-cross-modality-encoder | 24.76 | 39.0 | 55.4 | 82.26 | 74.0 |
| towards-vqa-models-that-can-read | - | - | 54.72 | - | - |