0
  • 我有 10 个折叠,每个折叠有 2 个文件夹face,并且background,我想在 train, validate 文件夹中复制不同的折叠,以便 train 或 validate 具有face不同background的折叠。

  • 我尝试了以下代码,但由于Google Colab使用python 3.6.9,我不能忽略文件夹存在的错误(如在python 3.9中),所以我收到以下错误:

错误:

try:
     mkdir(name, mode)
except OSError:
     # Cannot rely on checking for EEXIST, since the operating system

FileExistsError: [Errno 17] File exists: 'train/'

我的代码:

#--------------
# Split Dataset
#==============
# we Already have the test set
# !mkdir train
# !mkdir valid
import os, shutil

fileList = os.listdir("NewDataset")
fileList.sort()
for i in [x for x in range(10) if ((x != 9) and (x != 1))]:
    # print(fileList[i])
    # Train
    if(i < 7):
        subPathList = glob.glob('NewDataset/'+fileList[i]+'/**/', recursive=False)
        for subPath in subPathList:
            shutil.copytree(subPath, 'train/')

    # Validate
    else:
        subPathList = glob.glob('NewDataset/'+fileList[i]+'/**/', recursive=False)
        for subPath in subPathList:
            shutil.copytree(subPath, 'valid/')
4

1 回答 1

0

解决方案是递归地一个一个地复制文件:

import os, shutil

for i in [x for x in range(10) if ((x != 9) and (x != 1))]:
    # print(fileList[i])
    # Train
    if(i > 2):
        subPathList = glob.glob('/content/NewDataset/'+fileList[i]+'/**/', recursive=False)
        for subPath in subPathList:
            for im in os.listdir(subPath):
                imFullPath = os.path.join(subPath, im)
                targetPath = os.path.join('/content/train',subPath.split('/')[-2]+ '/')
                # print('Train: ', targetPath)
                shutil.copy(imFullPath, targetPath)
    # Validate
    else:
        subPathList = glob.glob('/content/NewDataset/'+fileList[i]+'/**/', recursive=False)
        for subPath in subPathList:
            for im in os.listdir(subPath):
                imFullPath = os.path.join(subPath, im)
                targetPath = os.path.join('/content/valid',subPath.split('/')[-2] + '/')
                # print('Validate: ', targetPath)
                shutil.copy(imFullPath, targetPath)
于 2020-10-29T18:15:24.070 回答