Visual Question Answering on GQA single-model (test)
91.2Binary AccuracyHuman
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| HumanModel setting=Single-model2019.07 | 91.2 | 87.4 | 98.4 | 98.9 | 97.2 | 89.3 | 89.3 | |
| NSMModel setting=Single-model, Strong supervision (functional programs)=false2019.07 | 78.94 | 49.25 | 93.25 | 96.41 | 84.28 | 3.71 | 63.17 | |
| DreamModel setting=Single-model2019.07 | 77.84 | 43.72 | 91.71 | 96.38 | 85.48 | 8.4 | 59.72 | |
| LXRTModel setting=Single-model2019.07 | 77.76 | 44.97 | 92.84 | 96.3 | 85.19 | 8.31 | 60.34 | |
| PVRModel setting=Single-model, Strong supervision (functional programs)=true2019.07 | 77.69 | 43.01 | 90.35 | 96.45 | 84.53 | 5.8 | 59.27 | |
| GRNModel setting=Single-model2019.07 | 77.53 | 43.35 | 88.63 | 96.18 | 84.71 | 6.06 | 59.37 | |
| SK T-BrainModel setting=Single-model, Strong supervision (functional programs)=true2019.07 | 77.42 | 43.1 | 90.78 | 96.26 | 85.27 | 7.54 | 59.19 | |
| MACModel setting=Single-model2019.07 | 71.23 | 38.91 | 81.59 | 96.16 | 84.48 | 5.34 | 54.06 | |
| BottomUpModel setting=Single-model2019.07 | 66.64 | 34.83 | 78.71 | 96.18 | 84.57 | 5.98 | 49.74 | |
| Lang+VisModel setting=Single-model2019.07 | 63.26 | 31.8 | 74.57 | 96.02 | 84.25 | 7.46 | 46.55 | |
| LanguageModel setting=Single-model2019.07 | 61.9 | 22.69 | 68.68 | 96.39 | 87.3 | 17.93 | 41.07 | |
| Local PriorModel setting=Single-model2019.07 | 47.9 | 16.66 | 54.04 | 84.33 | 84.31 | 13.98 | 31.24 | |
| Global PriorModel setting=Single-model2019.07 | 42.94 | 16.62 | 51.69 | 88.86 | 74.81 | 93.08 | 28.9 | |
| VisionModel setting=Single-model2019.07 | 36.05 | 1.74 | 62.4 | 35.78 | 34.84 | 19.99 | 17.82 |