PPOCRLabel标注识别图片自动裁剪

paddleocr自定义数据集标注图片，识别部分的图片自动裁剪的python代码。

import os
from PIL import Image
import json

list01 = []
with open("Label.txt", 'r', encoding="utf-8") as f:
    data = f.readlines()
    for item in data:
        pic_name = item.strip().split("t")[0].split("/")[1]
        json_info = item.strip().split("t")[1]
        # print(pic_name)
        # ok = json.dumps(pic_name)
        ok = json.loads(json_info)
        points_1 = ok[0]["points"]
        transcription_1 = ok[0]["transcription"]
        points_2 = ok[1]["points"]
        transcription_2 = ok[1]["transcription"]
        box01 = (round(points_1[0][0], 0), round(points_1[0][1], 0), round(points_1[1][0], 0), round(points_1[2][1], 0))
        # print(box01)
        box02 = (round(points_2[0][0], 0), round(points_2[0][1], 0), round(points_2[1][0], 0), round(points_2[2][1], 0))
        # print(box02)
        list01.append((pic_name, box01, box02, transcription_1, transcription_2))

pic_path = r"E:OurDatahandleP"
save_path = r"E:OurDatahandleR"
for dirpath, dirnames, filenames in os.walk(pic_path):
    for filename in filenames:
        print(filename)
        for item in list01:
            if item[0] == filename:
                img = Image.open(os.path.join(pic_path, filename))
                box01 = item[1]
                box02 = item[2]
                img01 = img.crop(box01)
                img02 = img.crop(box02)
                img01.save(os.path.join(save_path, filename.split(".")[0] + "_0_" + item[3] + ".jpg"))
                img02.save(os.path.join(save_path, filename.split(".")[0] + "_1_" + item[4] + ".jpg"))