Table 1. Video-based classification.
We can find that the proposed classification fusion model performs better than the single-mode recognition method in all aspects.
| Motion1 | Motion2 | Motion3 | Motion4 | Mean | |
|---|---|---|---|---|---|
| Pre | 95% | 90% | 90% | 95% | 92.5% |
| Rec | 95% | 90% | 85.7% | 100% | 92.7% |
| F1 | 95% | 90% | 87.8% | 97.4% | 92.6% |