python-拷贝图片/移除图片

随机划分训练集到测试集

最新推荐文章于 2025-09-17 10:50:50 发布

原创最新推荐文章于 2025-09-17 10:50:50 发布 · 1.5k 阅读

0 ·

CC 4.0 BY-SA版权

PyTorch 专栏收录该内容

25 篇文章

订阅专栏

该代码示例展示了如何从训练集文件夹(trainA/B/C)中随机选择494张图片，并将这些图片移动到测试集文件夹(testA/B/C)中，同时删除原训练集中的对应图片。

import numpy as np
import shutil
import os
root = '/root/……/datasets/face_newData'
trainA_path = root + '/trainA'
trainB_path = root + '/trainB'
trainC_path = root + '/trainC'

testA_path = root + '/testA'
testB_path = root + '/testB'
testC_path = root + '/testC'

ID = []
for i in range(494):
    idx = np.random.randint(10494)
    print idx
    if idx not in ID:
        ID.append(idx)
        A_path = trainA_path +  '/{}_A.jpg'.format(idx)
        A_new_path = testA_path + '/{}_A.jpg'.format(idx)
        shutil.copy(A_path, A_new_path)
        os.remove(A_path)


        B_path = trainB_path + '/{}_B.jpg'.format(idx)
        B_new_path = testB_path + '/{}_B.jpg'.format(idx)
        shutil.copy(B_path, B_new_path)
        os.remove(B_path)

        C_path = trainC_path + '/{}_C.jpg'.format(idx)
        C_new_path =testC_path + '/{}_C.jpg'.format(idx)
        shutil.copy(C_path, C_new_path)
        os.remove(C_path)
    else:
        continue

以上代码主要是从trainA/B/C中随机选取494张图片copy到testA/B/C中