Research Article
Using a Selective Ensemble Support Vector Machine to Fuse Multimodal Features for Human Action Recognition
Table 5
The comparison results between the proposed method and other machine learning methods.
| Dataset | Descriptor | SESVM (%) | Boosting (%) | Bagging (%) | SVM (%) | ANNs (%) |
| G3D | RGB-HOG | 83.7 | 83.2 | 75.2 | 80.2 | 74.2 | DMM-LBP | 83.2 | 82.4 | 79.4 | 83.4 | 80.2 | HJF | 83.7 | 84.0 | 80.5 | 83.6 | 74.4 | Mixed features | 91.7 | 89.5 | 83.2 | 87.7 | 82.4 |
| CAD60 | RGB-HOG | 85.3 | 87.3 | 80.0 | 82.0 | 76.2 | DMM-LBP | 86.0 | 87.4 | 79.2 | 84.3 | 80.5 | HJF | 88.6 | 89.2 | 80.2 | 84.4 | 74.2 | Mixed features | 91.8 | 89.1 | 83.3 | 87.6 | 82.2 |
|
|