[每日一氵]将分割的img和ann路径写到txt中 python 代码

这俩天，天天和分割的数据集打交道，和检测的label不一样，分割的label是一个mask，我习惯将他们写在一个txt中，形如：

img1_path ann1_path
img2_path ann2_path
img3_path ann3_path
img4_path ann4_path

每次都重写一遍，今儿索性放着这里，下次遇到直接copy过来改

import os
from pathlib import Path
from os import fspath

sep = " " # 一般 \t 更好
assert isinstance(sep, str)

train_list = []
train_img_root = "/root/baseline_dataset/img_dir/train"
for imgObj in Path(train_img_root).glob("**/*.jpg"):
    img_path = fspath(imgObj)
    ann_path = img_path.replace("img_dir", "ann_dir").replace("jpg", "png")
    assert os.path.exists(img_path), img_path
    assert os.path.exists(ann_path), ann_path
    
    temp = [img_path, ann_path]
    train_list.append(temp)

write_str = ""
train_txt = "/root/baseline_dataset/train.txt"

for item in train_list:
    write_str += sep.join(item) + "\n"
    
with open(train_txt, "w") as f:
    f.write(write_str)

print("文件写入完毕")

# 有个小bug, 就是文件的结尾多个 \n, 如果你读进来直接"split", 可能会有问题，建议读入前，先strip