Multiscale U-Net with Spatial Positional Attention for Retinal Vessel Segmentation

Figure 1

Global convolution network (a) and boundary refinement (b).

(a)

(b)