U-Net: Convolutional Networks for Biomedical Image Segmentation
OlafRonneberger,PhilippFischer,andThomasBroxComputerScienceDepartmentandBIOSSCentreforBiologicalSignallingStudies,UniversityofFreiburg,Germany摘要人们普遍认为,对深层网络的成功训练需要数千份已标注的训练样本。在本文中,我们提出了一个网络和训练策略,为了更有效