批量修改整理数据集

下面展示一些 内联代码片。
批量 整理

目标检测数据 批量整理

用于路径下各个文件夹内部图片和XML标签文件的同步批量重命名

路径部分需要小修改，主体部分可以直接套用

talk is cheap,show code

#coding:utf-8
from PIL import Image
import os
import glob
import xml.etree.ElementTree as ET
import xml.dom.minidom

# i = 1
dir_data = "/home/hs/backup/data/"
files = os.listdir(dir_data)
x = 0
for i in files:
    n =1
    x = x +1
    old_dir = dir_data + i
    new_dir = dir_data + "data_0903_" + str(x)
    tmp = i
    imgsdir = "/home/hs/backup/data/" + tmp + "/imgs"
    xmldir = "/home/hs/backup/data/" + tmp +"/xmls"

    newimg = "/home/hs/backup/data/" + tmp + "/img"
    newxml = "/home/hs/backup/data/" + tmp + "/xml"

    # imgsdir = "/home/hs/backup/data/data_0903_3/JPEGImages"
    # xmldir = "/home/hs/backup/data/data_0903_3/Annotations"

    # newimg = "/home/hs/backup/data/data_0903_3/img"
    # newxml = "/home/hs/backup/data/data_0903_3/xml"

    if not os.path.exists(newimg):
        os.mkdir(newimg)
    if not os.path.exists(newxml):
        os.mkdir(newxml)

    for xmlfile in os.listdir(xmldir):
        xmlname = os.path.splitext(xmlfile)[0]
        for pngfile in os.listdir(imgsdir):
            try:
                pngname = os.path.splitext(pngfile)[0]
                if pngname == xmlname:
                    # 修改图片文件名
                    # 图片文件名修改前后的路径
                    # if os.path.splitext(pngfile)[1] == '.png':
                    #     olddir = os.path.join(os.path.abspath(imgsdir), pngname + ".png")
                    #     newdir = os.path.join(os.path.abspath(newimg), str(i)+".png")
                    # elif os.path.splitext(pngfile)[1] == '.jpg':
                    #     olddir = os.path.join(os.path.abspath(imgsdir), pngname + ".jpg")
                    #     newdir = os.path.join(os.path.abspath(newimg), str(i)+".jpg")
                    # else:
                    #     olddir = os.path.join(os.path.abspath(imgsdir), pngname + ".JPG")
                    #     newdir = os.path.join(os.path.abspath(newimg), str(i)+".JPG")
                    if os.path.splitext(pngfile)[1] == '.png':
                        olddir = os.path.join(os.path.abspath(imgsdir), pngname + ".png")
                        newdir = os.path.join(os.path.abspath(newimg), '%05d' % int(n)+".png")
                    elif os.path.splitext(pngfile)[1] == '.jpg':
                        olddir = os.path.join(os.path.abspath(imgsdir), pngname + ".jpg")
                        newdir = os.path.join(os.path.abspath(newimg), '%05d' % int(n)+".jpg")
                    else:
                        olddir = os.path.join(os.path.abspath(imgsdir), pngname + ".JPG")
                        newdir = os.path.join(os.path.abspath(newimg), '%05d' % int(n)+".JPG")
                    # os.rename(olddir, newdir)
                    #print(xmlfile, '----->', str(i) + '.png')
                    # 修改filename结点属性
                    # 读取xml文件
                    dom = xml.dom.minidom.parse(os.path.join(xmldir, xmlfile))
                    root = dom.documentElement

                    # 获取标签对filename之间的值并赋予新值i
                    # root.getElementsByTagName('filename')[0].firstChild.data = str(i) + os.path.splitext(pngfile)[1]
                    root.getElementsByTagName('filename')[0].firstChild.data = '%05d' % int(n) + os.path.splitext(pngfile)[1]

                    # 将修改后的xml文件保存
                    # xml文件修改前后的路径
                    old_xmldir = os.path.join(xmldir, xmlfile)
                    # new_xmldir = os.path.join(newxml, str(i)+'.xml')
                    new_xmldir = os.path.join(newxml, '%05d' % int(n)+'.xml')
                    # 打开并写入
                    with open(old_xmldir, 'w') as fh:
                        dom.writexml(fh)
                    os.rename(old_xmldir, new_xmldir)
                    os.rename(olddir, newdir)
                    n += 1
            except Exception as e:
                pass
            continue
    os.rename(old_dir, new_dir)

    print('total number is ', n)