Image Super-Resolution Network Based on Feature Fusion Attention

Figure 4

Input and pixel attention channel weight (a). As we can see, the foreground is a desk with higher weight on the output (b).

(a) Example input image

(b) Output for PA channel