Video Question Answering on TGIF-QA original (test)
4.2825Repetition Count Loss (Mean L2)ST-VQA-Sp.
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ST-VQA-Sp.Attention mechanism=Spatial attention2017.04 | 4.2825 | 57.33 | 63.72 | 45.45 | |
| ST-VQA-ConcatInput feature representation=Concat (ResNet and C3D)2017.04 | 4.3759 | 60.13 | 65.7 | 48.2 | |
| ST-VQA-Tp.Attention mechanism=Temporal attention2017.04 | 4.3981 | 60.77 | 67.06 | 49.27 | |
| ST-VQA-C3DInput feature representation=C3D2017.04 | 4.4478 | 59.26 | 64.9 | 45.18 | |
| ST-VQA-ResNetInput feature representation=ResNet2017.04 | 4.5539 | 59.04 | 65.56 | 45.6 | |
| ST-VQA-Sp.Tp.Attention mechanism=Spatial and Temporal attention2017.04 | 4.5614 | 56.99 | 59.59 | 47.79 | |
| VIS+LSTMAggregation method=avg2017.04 | 4.8095 | 48.77 | 34.82 | 34.97 | |
| ST-VQA-TextInput feature representation=Text2017.04 | 5.0056 | 47.91 | 56.93 | 39.26 | |
| VIS+LSTMAggregation method=aggr2017.04 | 5.0921 | 46.84 | 56.85 | 34.59 | |
| Yu et al.2017.04 | 5.1387 | 56.14 | 63.95 | 39.64 | |
| VQA-MCBAggregation method=aggr2017.04 | 5.1738 | 58.85 | 24.27 | 25.7 | |
| VQA-MCBAggregation method=avg2017.04 | 5.5428 | 29.13 | 32.96 | 15.49 | |
| Random chance2017.04 | 6.9229 | 20 | 20 | 0.06 |