【YOLO】将多类别YOLO格式txt标签数据区分成单类别标签

        比如一个yolo格式的txt标签数据包含多个类别，现在需要将多类别区分训练，就需要将一个txt文件拆分成多个txt文件，python代码实现如下：

import os
import numpy as np
import shutil
img_path = '/train_images_labels_detection/labels/train'
fire_path= '/train_images_labels_detection/dataset1/labels/train'
smoke_path= '/train_images_labels_detection/dataset2/labels/train'

if not os.path.exists(fire_path):
    os.mkdir(fire_path)
if not os.path.exists(smoke_path):
    os.mkdir(smoke_path)

label_list = os.listdir(label_path)
for i in label_list:
    txtpath = os.path.join(label_path, i)
    with open(txtpath, 'r', encoding='utf-8') as file:
        fire = []
        smoke = []
        for line_num, line in enumerate(file, start=1):
            line = line.strip()
            if line:
                # 判断第一个字符,这里就列举了2类，此处可以根据自己数据集类别进行补充修改
                if line[0] == '0':
                    fire.append(line)
                else:
                    smoke.append(line)
    if len(fire)!=0:
        with open(os.path.join(fire_path,i), 'w', encoding='utf-8') as file1:
            for item in fire:
                file1.write(item + '\n')
    if len(smoke)!=0:
        with open(os.path.join(smoke_path,i), 'w', encoding='utf-8') as file2:
            for item in smoke:
                file2.write(item + '\n')

代码解释：

img_path：所有多类别标签数据地址，比如包含了火和烟2种类别，现在需要区分2者

fire_path：输出火的标签地址

smoke_path：输出烟的标签地址