Transferable Feature Representation for Visible-to-Infrared Cross-Dataset Human Action Recognition

Figure 1

The framework of cross-dataset feature alignment and generalization (CDFAG). Top: training. Bottom: testing.