数据清洗：批量删除xml长宽小于15*15的标注框

过小标注框可能是错误标注，或者对训练有干扰，运行下面python代码即可批量删除xml长宽小于15*15的标注框，标注框大小可自己调节。

一、删除xml长宽小于15*15的标注框

import os
import xml.etree.ElementTree as ET
from xml.dom.minidom import parse
from xml.etree.ElementTree import ElementTree,Element
import numpy as np

# file road
xml_ano = r'./Annotations'
xml_ano1 = r'./Annotations'
xml_list = os.listdir(xml_ano)

for xml_pa in xml_list:
    xml_path = xml_ano + '/' + xml_pa#xml文件的绝对路径
    tree = ET.parse(xml_path)
    root = tree.getroot()
    # print(root)
    nodes = root.findall('object')
    # print(nodes)

    for element1 in root.findall('object'):
        # print(element1)
        for element2 in element1.findall('bndbox'):
            xmin = element2.find('xmin').text
            ymin= element2.find('ymin').text
            xmax = element2.find('xmax').text
            ymax = element2.find('ymax').text
            mult = float(xmax) - float(xmin)
            mult1 = float(ymax) - float(ymin)
            if mult <=15 and mult1 <=15:
                # print(xml_path)
                root.remove(element1)
                print(xmin, ymin, xmax, ymax)

    tree.write(xml_ano1+'/' + xml_pa, encoding="utf-8",xml_declaration=True)

二、再删除所有空标注框的xml文件，运行下面python代码：

import xml.etree.ElementTree as ET
import os
 
# 批量删除整个文件夹所有的xml文件
def delete_all_xml(xml_path):
    filelist = os.listdir(xml_path)
    # 打开xml文档
    for xmlfile in filelist:
        doc = ET.parse(xml_path + xmlfile)
        root = doc.getroot()
        a = root.findall('object')
        if len(a) ==0:
            os.remove(xml_path+xmlfile)
 
if __name__ == '__main__':
    xml_path = "./Annotations/"
    delete_all_xml(xml_path)

三、再运行下面代码，同步删除空xml对应的图片即可：

import os
 
images_dir = './JPEGImages/'
xml_dir = './Annotations/'
 
#创建列表
xmls = []
#读取xml文件名(即：标注的图片名)
for xml in os.listdir(xml_dir):
    #xmls.append(os.path.splitext(xml)[0])    #append()参数：在列表末尾添加新的对象，即将所有文件名读入列表
    xmls.append(xml.split('.')[0])#splitext和split的区别：前者('0001','.jpg'), 后者('0001','jpg') 在此可选用
print(xmls)
 
#读取所有图片
for image_name in os.listdir(images_dir):
    image_name = image_name.split('.')[0]
    if image_name not in xmls:
        image_name = image_name + '.jpg'
        print(image_name)
        os.remove(os.path.join(images_dir,image_name))