Python实现随机划分图片数据集的示例代码

import os
import random
import shutil

def get_imlist(path):
    return [os.path.join(path, f) for f in os.listdir(path) if f.endswith('.jpg')]

def getData(src_path):

    dest_dir = src_path+'val' #划分出来的验证集
    if not os.path.isdir(dest_dir):
        os.mkdir(dest_dir)

    img_list = get_imlist(src_path)
    random.shuffle(img_list)
    le = int(len(img_list) * 0.8)  # 这个可以修改划分比例
    for f in img_list[le:]:
        shutil.move(f, dest_dir)


'''
函数功能：
划分数据集
'''
def SplitImg(filePath):
    getData(filePath)

'''
函数功能：
根据划分的数据集进行移动标注文件
'''
def MoveAn(filePathAn,filePathImg):
    Imgs=os.listdir(filePathImg)
    if not os.path.isdir(filePathAn+'val'):
        os.mkdir(filePathAn+'val')

    for file in os.listdir(filePathAn):
        #print(filePathAn,filePathImg)
        #print(os.path.join(filePathAn,file),os.path.join(filePathAn+'val',file))
        if file[:-4]+'.jpg' in Imgs:

            shutil.move(os.path.join(filePathAn,file),os.path.join(filePathAn+'val',file))

if __name__=='__main__':
    filePath='./宠物/宠物'# 换成你的数据集

    #拆分的数据集
    SplitImg(filePath)
    filePathAn='./宠物/宠物An'# 换成你的标注文件地址

    # 根据数据集进行移动标注文件
    MoveAn(filePathAn,filePath+'val')

import os
import random
import shutil
 

def moveFile(input1,input2,save1,save2):
    pathDir = os.listdir(input1)  # 取图片的原始路径
    random.seed(1)
    filenumber = len(pathDir)  # 原文件个数
    rate = 0.1  # 抽取的验证集的比例，占总数据的多少
    picknumber = int(filenumber * rate)  # 按照rate比例从文件夹中取一定数量图片
    sample = random.sample(pathDir, picknumber)  # 随机选取需要数量的样本图片
    print(sample)
    list_len=len(sample)
    print(list_len)
    list=[]
    for i in range(len(sample)):
        list.append(sample[i].split('.')[0])
    print(list)
    for flie_name in list:
        path_img=os.path.join(input1,flie_name+'.jpg')
        shutil.move(path_img,save1)
        path_lab=os.path.join(input2,flie_name+'.txt')
        shutil.move(path_lab,save2)
 
if __name__ == '__main__':
    input_path1='./train1200/images/train'
    input_path2= './train1200/labels/train'
    save_img='./train1200/images/val'
    save_lab='./train1200/labels/val'
    if not os.path.exists(save_lab):
        os.makedirs(save_lab)
    if not os.path.exists(save_img):
        os.makedirs(save_img)
    moveFile(input_path1,input_path2,save_img,save_lab)

from sklearn.model_selection import train_test_split

#X是所有的样本特征，y是目标变量，test_size是测试集占总样本数的比例
# random_state是随机数发生器的种子，保证每次划分结果一致
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0,3, random_state=42)

import random

def split(fname, train_ratio, var_ratio):
    lines = fname.readlines()
    n_total = len(lines)          # 获取数据集的总长度

    train_offset = int(n_total * train_ratio)
    val_offset = int(n_total * (train_ratio + var_ratio))
    random.shuffle(fname.read())  # 按行打乱顺序

    train_data = open('train.txt.bio', 'wb')
    val_data = open('val.txt.bio', 'wb')
    test_data = open('test.txt.bio', 'wb')


    for i, line in enumerate(lines):
        if i < train_offset:
            train_data.write(line)
        elif i < val_offset:
            val_data.write(line)
        else:
            test_data.write(line)

    train_data.close()
    val_data.close()
    test_data.close()


if __name__ == "__main__":
   fname = open('en/en_total.txt.bio', "rb")
   split(fname, train_ratio = 0.6, var_ratio = 0.2)
   fname.close()