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

Figure 7

The confusion matrix based on DMM-LBP features on the G3D dataset.