划分数据的方法

1、分割数据集（图片）进行存放

首先整体目录是这样的

以下是运行代码的结果

分割后的图片名保持一致

下面是代码（可能比较繁琐，并没有进行优化）

import os
import random
from PIL import Image
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

def division_train_eval(data_path):
    list = os.listdir(data_path)
    for i in range(0, len(list)):
        train_path_temporary = './data/train/'+ str(list[i])
        if not os.path.exists(train_path_temporary):
            os.makedirs(train_path_temporary)
        eval_path_temporary = './data/eval/'+ str(list[i])
        if not os.path.exists(eval_path_temporary):
            os.makedirs(eval_path_temporary)
        data_list = []
        path = os.path.join(data_path, list[i])
        if os.path.isdir(path):  # 判断是否是文件夹
            for j in os.listdir(path):
                path_class = path + "\\" + str(j)
                data_list.append(path_class)
        random.shuffle(data_list)
        print("==================================")
        print(len(data_list))
        train_list = []
        eval_list = []
        for i in range(int(len(data_list) * 0.8)):
            train_list.append(data_list[i])
        for i in range(int(len(data_list) * 0.8), len(data_list)):
            eval_list.append(data_list[i])
        print(len(train_list))
        print(len(eval_list))
        for i in train_list:
            I = Image.open(i)
            path_name = i.split('\\')[-1]
            path_save = train_path_temporary + '\\' +path_name
            I.save(path_save)
        for i in eval_list:
            I = Image.open(i)
            path_name = i.split('\\')[-1]
            path_save = eval_path_temporary + '\\' +path_name
            I.save(path_save)
if __name__ == '__main__':
    data_path = './data/dataset'
    division_train_eval(data_path)