python将一组图篇集目录中随机抽取样本进行测试

想将图片集目录中的每个目录随机抽取X个样本用于测式，并重以md5命名文件。python代码如下

import os, random, shutil
import hashlib
def rand_sampling(pathDir, n):
    """
    :param number_set: 数字列表
    :param n: 组合数位数
    :return: 组合数
    """

    result = []
    for i in range(1, n+1):
        pic = random.choice(pathDir)
        result.append(pic)
        pathDir.remove(pic)

    return result
def file_md5(file_name ):
    if not os.path.isfile(file_name):
        return
    with open(file_name, 'rb') as fp:
        data = fp.read()

    return  hashlib.md5(data).hexdigest()
def moveFile(source,aim,picknumber):
    pathDir = os.listdir(source)    #取图片的原始路径

    if len(pathDir)<= picknumber:
        sample = pathDir
    else:
        sample = rand_sampling(pathDir, picknumber)
    print(source)
    print("len:",len(pathDir))
    for name in sample:
            s = os.path.join(source, name)

            a = os.path.join(aim, name)
            md5_name=file_md5(s)+".jpg"
            #shutil.copyfile(source+name, aim+name)
            shutil.copyfile(source + name, aim + md5_name)
    return

if __name__ == '__main__':
    picknumber = 10 #每次抽取的数量
    fileDir = r"E:\pictures\data\for_hegui"    #源图片文件夹路径
    tarDir = r'E:\测试图集\随机抽取'    #移动到新的文件夹路径
    shutil.rmtree(tarDir)  #清空新的文件夹路径

    for root, dirs, files in os.walk(fileDir):
        for d in dirs:
            source = os.path.join(fileDir, d) +'/'
            aim = os.path.join(tarDir, d)+'/'
            if not os.path.exists(aim):
                os.makedirs(aim)
            moveFile(source,aim,picknumber)

可以保证每次抽取数量

下面代码是抽取后，自动将英文文件夹改名为中文的，便于人工测试，只要在Dictionary_chineseDirs 中写好对应关键即可。

import os, random, shutil
import hashlib
def rand_sampling(pathDir, n):
    """
    :param number_set: 数字列表
    :param n: 组合数位数
    :return: 组合数
    """

    result = []
    for i in range(1, n+1):
        pic = random.choice(pathDir)
        result.append(pic)
        pathDir.remove(pic)

    return result
def file_md5(file_name ):
    if not os.path.isfile(file_name):
        return
    with open(file_name, 'rb') as fp:
        data = fp.read()

    return  hashlib.md5(data).hexdigest()
def moveFile(source,aim,picknumber):
    pathDir = os.listdir(source)    #取图片的原始路径

    if len(pathDir)<= picknumber:
        sample = pathDir
    else:
        sample = rand_sampling(pathDir, picknumber)
    print(source)
    print("len:",len(pathDir))
    for name in sample:
            s = os.path.join(source, name)

            a = os.path.join(aim, name)
            md5_name=file_md5(s)+".jpg"
            #shutil.copyfile(source+name, aim+name)
            shutil.copyfile(source + name, aim + md5_name)
    return

if __name__ == '__main__':
    picknumber = 50 #每次抽取的数量
    fileDir = r"E:\pictures\data\for_hegui"    #源图片文件夹路径
    tarDir = r'E:\测试图集\随机抽取'    #移动到新的文件夹路径

    if not os.path.exists(tarDir):
        os.makedirs(tarDir)
    else:
        shutil.rmtree(tarDir)  # 清空新的文件夹路径

    Dictionary_chineseDirs = {
                            
                            'document':'文档',
                            'drawings':'一般卡通',
                            'group_photo':'集体照',                         
                            'inform':'广告海报',
                            'neutral':'正常',
                            'people':'人像',
                            'redheadfile':'红头文件',                            
                            'suit':'西装',
                   }
    for root, dirs, files in os.walk(fileDir):
        for d in dirs:
            source = os.path.join(fileDir, d) +'/'

            if( d  in Dictionary_chineseDirs.keys()):
                chineseDir=Dictionary_chineseDirs[d]
            else:
                chineseDir=d
            #aim = os.path.join(tarDir, d)+'/'
            aim = os.path.join(tarDir, chineseDir) + '/'
            if not os.path.exists(aim):
                os.makedirs(aim)
            moveFile(source,aim,picknumber)