Visual Question Answering on CLEVR pixels (test)
95.5Overall AccuracyCNN+LSTM+RN
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CNN+LSTM+RN2017.06 | 95.5 | 90.1 | 97.8 | 93.6 | 97.9 | 97.1 | |
| Human2017.06 | 92.6 | 86.7 | 96.6 | 86.5 | 95 | 96 | |
| CNN+LSTM+SAimplementation=Authors' implementation, hyperparameters=optimized, training=fully end-to-end2017.06 | 76.6 | 64.4 | 82.7 | 77.4 | 82.6 | 75.4 | |
| CNN+LSTM+SA2017.06 | 68.5 | 52.2 | 71.1 | 73.5 | 85.3 | 52.3 | |
| CNN+LSTM2017.06 | 52.3 | 43.7 | 65.2 | 67.1 | 49.3 | 53 | |
| LSTM2017.06 | 46.8 | 41.7 | 61.1 | 69.8 | 36.8 | 51.8 | |
| Q-type baseline2017.06 | 41.8 | 34.6 | 50.2 | 51 | 36 | 51.3 |