Python实现数据集划分(训练集和测试集)

前面是分部讲解，完整代码在最后。

导入模块 ：

import os
from shutil import copy, rmtree
import random

创建文件夹 ：

def make_file(file_path: str):
   if os.path.exists(file_path):
       rmtree(file_path)
   os.makedirs(file_path)

划分数据集的比例，本文是0.1：验证集的数量占总数据集的10％比如填0.1就是验证集的数量占总数据集的10％

random.seed(0)  
split_rate = 0.1

数据集的存放：新建一个数据文件夹，将划分的数据集存放进去

data_path = r'D:\chengxu\data\caodi'  # 数据集存放的地方
data_root = r'D:\chengxu\data\cd'  # 这里是生成的训练集和验证集所处的位置，这里设置的是在当前文件夹下。
data_class = [cla for cla in os.listdir(data_path)]
print("数据的种类分别为：")
print(data_class)  # 输出数据种类

建立训练集文件夹：

train_data_root = os.path.join(data_root, "train")  # 训练集的文件夹名称为 train
make_file(train_data_root)
for num_class in data_class:
   make_file(os.path.join(train_data_root, num_class))

建立测试集文件夹：

val_data_root = os.path.join(data_root, "val")  # 验证集的文件夹名称为 val
make_file(val_data_root)
for num_class in data_class:
   make_file(os.path.join(val_data_root, num_class))

划分数据：

for num_class in data_class:
   num_class_path = os.path.join(data_path, num_class)
   images = os.listdir(num_class_path)
   num = len(images)
   val_index = random.sample(images, k=int(num * split_rate))  # 随机抽取图片
   for index, image in enumerate(images):
       if image in val_index:
           # 将划分到验证集中的文件复制到相应目录
           data_image_path = os.path.join(num_class_path, image)
           val_new_path = os.path.join(val_data_root, num_class)
           copy(data_image_path, val_new_path)
       else:
           # 将划分到训练集中的文件复制到相应目录
           data_image_path = os.path.join(num_class_path, image)
           train_new_path = os.path.join(train_data_root, num_class)
           copy(data_image_path, train_new_path)
   print("\r[{}] split_rating [{}/{}]".format(num_class, index + 1, num), end="")  # processing bar
   print()
print("       ")
print("       ")
print("划分成功")

完整代码：

import os
from shutil import copy, rmtree
import random

def make_file(file_path: str):
   if os.path.exists(file_path):

rmtree(file_path)
   os.makedirs(file_path)

random.seed(0)

# 将数据集中10％的数据划分到验证集中
split_rate = 0.1  
data_path = r'D:\chengxu\data\caodi'  # 数据集存放的地方，建议在程序所在的文件夹下新建一个data文件夹，将需要划分的数据集存放进去
data_root = r'D:\chengxu\data\cd'  # 这里是生成的训练集和验证集所处的位置，这里设置的是在当前文件夹下。

data_class = [cla for cla in os.listdir(data_path)]
print("数据的种类分别为：")
print(data_class)  
# 建立保存训练集的文件夹
train_data_root = os.path.join(data_root, "train")  # 训练集的文件夹名称为 train
make_file(train_data_root)
for num_class in data_class:
   # 建立每个类别对应的文件夹
   make_file(os.path.join(train_data_root, num_class))

# 建立保存验证集的文件夹
val_data_root = os.path.join(data_root, "val")  # 验证集的文件夹名称为 val
make_file(val_data_root)
for num_class in data_class:
   # 建立每个类别对应的文件夹
   make_file(os.path.join(val_data_root, num_class))

for num_class in data_class:
   num_class_path = os.path.join(data_path, num_class)
   images = os.listdir(num_class_path)
   num = len(images)

val_index = random.sample(images, k=int(num * split_rate))  # 随机抽取图片
   for index, image in enumerate(images):
       if image in val_index:

data_image_path = os.path.join(num_class_path, image)
           val_new_path = os.path.join(val_data_root, num_class)
           copy(data_image_path, val_new_path)
       else:

data_image_path = os.path.join(num_class_path, image)
           train_new_path = os.path.join(train_data_root, num_class)
           copy(data_image_path, train_new_path)
   print("\r[{}] split_rating [{}/{}]".format(num_class, index + 1, num), end="")  # processing bar
   print()

print("       ")
print("       ")
print("划分成功")

来源：https://blog.csdn.net/m0_61497715/article/details/129860228