Randomly assign training set, test set and validation set

This code is more suitable for a dataset that corresponds to a labeled file for a picture. After random allocation, all files are divided into train, test, val and the corresponding label data will also be separated.

It's easier to do if you just have images without annotation data. This article will not repeat them.

The following is the code, the core part is to use a random sort.

import os  # os是用来切换路径和创建文件夹的。
import random
from shutil import copy  # shutil 是用来复制黏贴文件的

# "blossom_end_rot", "graymold", "powdery_mildew", "spider_mite", "spotting_disease"

img_path = "/Users/Dong/Desktop/dataPath/data"
labels_path = "/Users/Dong/Desktop/dataPath/labels"

filelist_img_name = []
filelist_label_name = []

for root, dir, filenames in os.walk(img_path):
    for filename in filenames:
        if '.jpg' in filename:
            file_path = os.path.join(root, filename)
            print(file_path)
            filelist_img_name.append(file_path)
        else:
            pass
            # print("it is not a jpg file.", filename)

# .jpg.mat是我的标注文件的后缀名，您修改为自己的xml或者json或者其他类型即可。
for root, dir, filenames in os.walk(labels_path):
    for filename in filenames:
        if '.jpg.mat' in filename:
            filelist_label_name.append(filename.split('.jpg.mat')[0])
        else:
            pass
            # print("it is not a label file.", filename)

print(len(filelist_img_name))
print(len(filelist_label_name))
# 我一共含有580张图片，因此 348 ，464，464-580这几个数字是我自己计算的。我的比例是60% 20% 20%
# 如果您有10000张图片和10000张标注，那么您可以设置为6000，8000，8000:，以此类推
# 我没有写自动创建文件夹的那部分代码，因此你要自己首先把相关文件夹创建好。
random.shuffle(filelist_img_name)
filelist_img_name = filelist_img_name
for tr in filelist_img_name[0:348]:
    copy(tr, '/Users/Dong/Desktop/dataPath_new/data/train/')
    tr = tr.replace('data/', 'labels/')
    tr = tr.replace('.jpg', '.jpg.mat')
    copy(tr, '/Users/Dong/Desktop/dataPath_new/labels/train/')

for val in filelist_img_name[348:464]:
    copy(val, '/Users/Dong/Desktop/dataPath_new/data/val/')
    val = val.replace('data/', 'labels/')
    val = val.replace('.jpg', '.jpg.mat')
    copy(val, '/Users/Dong/Desktop/dataPath_new/labels/val/')

for test in filelist_img_name[464:]:
    copy(test, '/Users/Dong/Desktop/dataPath_new/data/test/')
    test = test.replace('data/', 'labels/')
    test = test.replace('.jpg', '.jpg.mat')
    copy(test, '/Users/Dong/Desktop/dataPath_new/labels/test/')