将数据集简单地划分为测试集以及训练集
import os
import random
train_percent = .9 # 训练集所占比例
valid_percent = 0.1 # 测试集所占比例
xmlfilepath = 'data-delete' #图片所在地相对路径
total_file = os.listdir(xmlfilepath) #以列表地形式将文件夹中的图片名字罗列出
num = len(total_file)
list = range(num)
tr = int(num * train_percent) #训练集的个数
tval = int(num * valid_percent)
trainval = random.sample(list, tval) #在list里面随机抽取 tval个数重新构成一个列表,当作验证集
train = random.sample(list, tr)
ftrain = open('train.txt', 'w')
fval = open('valid.txt', 'w')
for i in list:
name = "D:\yolov4-zhaban\data\data-delete/" + total_file[i] + '\n'
if i in trainval:
fval.write(name)
else:
ftrain.write(name)
ftrain.close()
fval.close() #啊创建完毕,关闭文件