基于 UnetPlusPlus 网络对人体轮廓分割实战【包含数据集、完整代码、训练好的结果、权重文件等等】

[train hyper-parameters: Namespace(batch_size=8, epochs=5, image_size=(160, 240), img_f='.jpeg', lr=0.001, lr_decay='lr_step', lr_step_epoch=(2, 3), lr_step_gamma=0.1, lrf=0.001, mask_f='.png', optim='RMSProp', threshold=0)] epoch,train loss,train dice,val loss,val dice 1,0.5688067771430708,0.3867054187172688,0.48088109130039813,0.6246314460318536 2,0.4877991615197597,0.6156941926122731,0.5068286066502332,0.5873675088302116 3,0.4549131150174345,0.654489684308696,0.45422672856599094,0.6527850328654131 4,0.4457938159123445,0.6633376516083367,0.4499694495275617,0.6550341106468477 5,0.4419038074138837,0.6695258481125547,0.4489714052155614,0.6552382077334914