Channel-Wise Spatiotemporal Aggregation Technology for Face Video Forensics

Figure 2

The architecture of the CWSA module; a colorful rectangle denotes a feature map output from the backbone.