search

zz10001 / LITS2017-main2

liver tumor segmentation(3d UNet) for LITS2017
25 stars 7 forks source link

取训练集与验证集时的问题 #4

Closed Jiangj512 closed 2 years ago

Jiangj512 commented 2 years ago

train_img_paths, val_img_paths, train_mask_paths, val_mask_paths = train_test_split(img_paths, mask_paths, test_size=0.3, random_state=39) 这样取训练集与验证集不会导致有交叉吗，来自同一个ct的不同切块同时被分到训练集与验证集中。

zz10001 commented 2 years ago

会，所以训练时候验证集的指标会很高，你可以尝试下处理这个问题，项目不在维护了