Using a Selective Ensemble Support Vector Machine to Fuse Multimodal Features for Human Action Recognition

Figure 8

The confusion matrix based on HJF on the G3D dataset.