Fig. 4From: PyConvU-Net: a lightweight and multiscale network for biomedical image segmentationSegmentation comparisons. From left to right, the columns represent the original image, mask, U-Net predictions, U-Net++ predictions, Resnet34_UNet predictions, FCN8s predictions, and PyConvU-Net predictions respectively. The red curve shows the actual area of the organ. The markers of the last row indicate the key areaBack to article page