Table 4.
Ablation experiments on Swin-Transformer.
| Method | Dataset | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Swimming Technique Datasets | ImageNet Datasets | Kinetics Datasets | Sports-1M Dataset | |||||||||||||
| Parameters(M) | Flops(G) | Inference Time(ms) | Trainning Time(s) | Parameters(M) | Flops(G) | Inference Time(ms) | Trainning Time(s) | Parameters(M) | Flops(G) | Inference Time(ms) | Trainning Time(s) | Parameters(M) | Flops(G) | Inference Time(ms) | Trainning Time(s) | |
| MCNN | 328.21 | 304.96 | 249.11 | 303.41 | 290.37 | 250.07 | 331.06 | 282.68 | 341.81 | 257.36 | 318.86 | 284.97 | 296.08 | 208.04 | 249.22 | 282.81 |
| MRNN | 259.14 | 246.81 | 275.01 | 231.44 | 284.81 | 396.08 | 249.56 | 215.22 | 399.37 | 306.77 | 335.97 | 330.68 | 399.67 | 338.17 | 287.30 | 262.95 |
| MGAN | 269.87 | 338.84 | 376.72 | 389.20 | 229.19 | 360.81 | 279.19 | 305.23 | 255.55 | 257.07 | 341.24 | 351.83 | 379.39 | 391.93 | 335.80 | 370.03 |
| Ours | 107.88 | 121.12 | 158.02 | 213.89 | 137.65 | 168.31 | 222.03 | 154.32 | 207.48 | 103.57 | 108.22 | 157.11 | 151.60 | 201.48 | 211.11 | 104.16 |