图片标注处理程序

1、A文件夹中有图片和xml文件，图片中的标准不止一个类别，记为aa、bb、cc

通过这个程序，把A文件夹中所有标注为aa目标的截图、bb目标的截图、cc目标的截图都分类放到B文件夹中

import xml.etree.ElementTree as ET
import glob
import os
from PIL import Image


PATH_TO_ANNOTATIONS = r'C:/Users/Administrator/Desktop/A'
PATH_TO_IMAGES = r'C:/Users/Administrator/Desktop/A'
PATH_TO_SAVE = r'C:/Users/Administrator/Desktop/B'


def main():
    xml_files = glob.glob(os.path.join(PATH_TO_ANNOTATIONS, "*.xml"))
    index = 0
    for xml_file in xml_files:
        print("Progress:%d/%d" %(index + 1, len(xml_files)))
        print(xml_file)
        index = index + 1
        # 解析xml文件
        tree = ET.parse(xml_file)
        annotation = tree.getroot()
        obj_num = 0
        img = None
        for obj in annotation.findall('object'):
            value = (annotation.find('filename').text+'.jpg',
                     int(annotation.find('size').find('width').text),
                     int(annotation.find('size').find('height').text),
                     obj.find('name').text,
                     int(obj.find('bndbox').find('xmin').text),
                     int(obj.find('bndbox').find('ymin').text),
                     int(obj.find('bndbox').find('xmax').text),
                     int(obj.find('bndbox').find('ymax').text)
                     )
            # 获取对应的图片文件
            image_file = xml_file.split('\')[-1].split('.')[0]
            crop_file = os.path.join(PATH_TO_SAVE, value[3], image_file + '_%d.jpg' % obj_num)
            image_file = os.path.join(PATH_TO_IMAGES, image_file + '.' + value[0].split('.')[-1])

            # 提取图片
            if obj_num == 0:
                img = Image.open(image_file)
            crop_img = img.crop((min(value[4], value[6]), min(value[5], value[7]), max(value[4], value[6]), max(value[5], value[7])))
            if not os.path.exists(os.path.join(PATH_TO_SAVE, value[3])):
                os.makedirs(os.path.join(PATH_TO_SAVE, value[3]))
            crop_img.save(crop_file)
            obj_num = obj_num + 1

if __name__ == '__main__':
    main()

2、A文件夹中有图片和xml文件，有些图片中没有目标，就不用标注，因此就没有xml文件。把A文件夹中有xml的图片以及对应的xml文件提取出来，复制到B文件夹中

import os
from PIL import Image
import xml.dom.minidom

src_file_path = "C:/Users/Administrator/Desktop/1113/A"
dst_file_path = "C:/Users/Administrator/Desktop/1113/B"



dir_list=os.listdir(src_file_path)


for file in dir_list:
    if file.endswith('.xml'):
        str_=file.split(".xml")[0]
        print(str_)
        img_src_path = src_file_path + '/' + str_+'.jpg'
        im = Image.open(img_src_path)

        img_save_path =dst_file_path+'/'+str_+'.jpg'
        im.save(img_save_path)

        xml_src_path=src_file_path+'/'+file
        dom=xml.dom.minidom.parse(xml_src_path)
        xml_save_path=dst_file_path+'/'+file
        with open(xml_save_path, 'w') as fh:
            dom.writexml(fh)