RGB-D Human Action Recognition of Deep Feature Enhancement and Fusion Using Two-Stream ConvNet

Figure 9

V-Stream loss curve of training and testing.