Action Recognition on UAV-Human 20
47.4Top-1 AccuracyAZTR
Evaluation results
| Method | Links | |
|---|---|---|
| AZTRBackbone=X3D-M, Frames=16, Input Size=224 x 224, Init.=Kinetics2023.03 | 47.4 | |
| AZTRBackbone=X3D-M, Frames=16, Input Size=224 x 224, Init.=None2023.03 | 39.2 | |
| FARBackbone=X3D-M, Frames=8, Input Size=620 x 620, Init.=Kinetics2023.03 | 39.1 | |
| FARBackbone=X3D-M, Frames=16, Input Size=224 x 224, Init.=Kinetics2023.03 | 31.9 | |
| X3D-MBackbone=X3D-M, Frames=16, Input Size=224 x 224, Init.=Kinetics2023.03 | 30.6 | |
| FARBackbone=I3D, Frames=8, Input Size=540 x 960, Init.=Kinetics2023.03 | 29.21 | |
| FARBackbone=X3D-M, Frames=8, Input Size=540 x 540, Init.=None2023.03 | 28.8 | |
| X3D-LBackbone=X3D-L, Frames=16, Input Size=224 x 224, Init.=None2023.03 | 27.6 | |
| FARBackbone=X3D-M, Frames=16, Input Size=224 x 224, Init.=None2023.03 | 27.6 | |
| X3D-MBackbone=X3D-M, Frames=16, Input Size=224 x 224, Init.=None2023.03 | 27 | |
| FNetBackbone=I3D, Frames=8, Input Size=540 x 960, Init.=Kinetics2023.03 | 24.39 | |
| X3D-SBackbone=X3D-S, Frames=16, Input Size=224 x 224, Init.=None2023.03 | 21.5 | |
| I3DBackbone=ResNet-101, Frames=8, Input Size=540 x 960, Init.=Kinetics2023.03 | 21.06 |