Adaptive Panoramic Video Multicast Streaming with Limited FoV Feedback

Figure 5

The saliency maps of only one frame are shown for each selected video. The prediction map is derived by combining the saliency map with the ground-truth user viewport.