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

Figure 3

Illustration of aligned-to-generalized encoders generalizing aligned features across visible light and infrared action datasets.