Dive Recognition on MTL-AQA (test)
96.32Position AccuracyC3D-AVG
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| C3D-AVGCNN Type=3D, Number of Frames=962019.12 | 96.32 | 99.72 | 97.45 | 96.88 | 93.2 | |
| HalluciNet (ResNet-18)CNN Type=2D, Number of Frames=62019.12 | 91.78 | 99.43 | 95.47 | 88.1 | 89.24 | |
| VGG11CNN Type=2D, Number of Frames=62019.12 | 90.08 | 99.43 | 92.07 | 83 | 86.69 | |
| HalluciNet (VGG11)CNN Type=2D, Number of Frames=62019.12 | 89.52 | 99.43 | 96.32 | 86.12 | 88.1 | |
| MSCADCCNN Type=3D, Number of Frames=162019.12 | 78.47 | 97.45 | 84.7 | 76.2 | 82.72 | |
| Nibali et al.CNN Type=3D, Number of Frames=162019.12 | 74.79 | 98.3 | 78.75 | 77.34 | 79.89 |