Fig. 2
From: Semantic segmentation to identify bladder layers from H&E Images
Modified U-Net architecture with 12 layers. Numbers on the top of blocks represent the number of feature maps, while numbers on the slide of blocks represent the feature map size