统计XML标签种类和个数

# -*- coding: utf-8 -*-

"""
Created on 20-3-6

@author: zmcldw

@requirements: PyCharm 2018.2.4; Python 3.6.0 |Anaconda 4.3.1 (64-bit)

@decription: statistic the numbers of classes in labels
"""


import os
import xml.dom.minidom


AnnoPath = ''
Annolist = os.listdir(AnnoPath)
rate = {} # 创建一个字典用于存放标签名和对应的出现次数
# total = 0
for annotation in Annolist:
    fullname = AnnoPath + annotation
    dom = xml.dom.minidom.parse(fullname) # 打开XML文件
    collection = dom.documentElement # 获取元素对象
    objectlist = collection.getElementsByTagName('object') # 获取标签名为object的信息
    for object in objectlist:
        namelist = object.getElementsByTagName('name') # 获取子标签name的信息
        objectname = namelist[0].childNodes[0].data # 取到name具体的值
        if objectname not in rate: # 判断字典里有没有标签，如无添加相应字段
            rate[objectname] = 0
        rate[objectname] += 1
        # total += 1

print(rate)
# print(total)

# 画图
# import matplotlib.pyplot as plt
# object = []
# number = []
# for key in rate:
#     object.append(key)
#     number.append(rate[key])
# plt.figure()
# plt.bar(object, number)
# plt.title('result')
# plt.show()