VOC格式数据集xml文件处理-类别增、删、改工具
登录 / 注册
标签搜索
侧边栏壁纸
博主昵称
jupiter

  • 累计撰写 331 篇文章
  • 累计收到 63 条评论
数据集和工具

VOC格式数据集xml文件处理-类别增、删、改工具

jupiter
jupiter
2022-03-14 / 0 评论 / 572 阅读 / 正在检测是否收录...
温馨提示：
本文最后更新于2022年04月01日，已超过754天没有更新，若内容或图片失效，请留言反馈。

1.删除xml中某一类别

import xmltodict
import os
from xml.dom import minidom
from tqdm import tqdm

# filter class
class_filter_list = ["person","baggage car"]

xml_dir="./新建文件夹/"# VOC xml文件所在文件夹

# 加入进度条支持
pbar = tqdm(total=len(os.listdir(xml_dir)))
pbar.set_description("VOC xml filter") # 设置前缀

# 逐一处理xml文件
for file in os.listdir(xml_dir):
    if not file.endswith(".xml"):
        pbar.update(1)
        continue # 过滤掉非xml文件
        
    xml_path = os.path.join(xml_dir,file) # 拼接xml地址
    
    # 读取xml文件+转为字典
    with open(xml_path,'r',encoding="utf8") as f:
        xml_str = f.read()
    xml_dic = xmltodict.parse(xml_str)
    
    # 删除掉待过滤的类别
    tmp_obj_list = xml_dic["annotation"]["object"]
    tmp_obj_list_fiter = []
    for tmp_obj in tmp_obj_list:
        if not tmp_obj["name"] in class_filter_list:
            tmp_obj_list_fiter.append(tmp_obj)
    xml_dic["annotation"]["object"] = tmp_obj_list_fiter
    xmlstr = xmltodict.unparse(xml_dic)
    xml_pretty_str = minidom.parseString(xmlstr).toprettyxml()
    
    with open(xml_path,"w",encoding="utf8") as f:
        f.write(xml_pretty_str)
    
    pbar.update(1)
pbar.close()

2.指定帧区间添加特定bbox

# 在固定位置添加指定的候选框

import xmltodict
import os
from xml.dom import minidom
from tqdm import tqdm
import copy

xml_dir="./data_handle/xml/"# 待处理xml文件夹

# 待添加bbox
xml = """
    <object>
        <name>LuggageVehicle</name>
        <pose>Unspecified</pose>
        <truncated>0</truncated>
        <difficult>0</difficult>
        <bndbox>
            <xmin>212</xmin>
            <ymin>71</ymin>
            <xmax>284</xmax>
            <ymax>110</ymax>
        </bndbox>
    </object>
"""

xml_dict = xmltodict.parse(xml)
obj = xml_dict["object"]
cls_add = obj["name"]
bndbox = obj["bndbox"]
bbox_add = [int(bndbox["xmin"]),int(bndbox["ymin"]),int(bndbox["xmax"]),int(bndbox["ymax"])]

# 起止帧id
frame_begin = 84543
frame_end = 84919

# 加入进度条支持
pbar = tqdm(total=len(os.listdir(xml_dir)))
pbar.set_description("VOC xml add class") # 设置前缀

# 逐一处理xml文件
for file in os.listdir(xml_dir):
    if not file.endswith(".xml"):
        pbar.update(1)
        continue # 过滤掉非xml文件
    
    if int(file[:-4])<frame_begin or int(file[:-4])>frame_end:
        pbar.update(1)
        continue
        
    xml_path = os.path.join(xml_dir,file) # 拼接地址
    
    # 读取xml文件+转为字典
    with open(xml_path,'r',encoding="utf8") as f:
        xml_str = f.read()
    xml_dic = xmltodict.parse(xml_str)
    
    # 添加类别
    tmp_obj_list = xml_dic["annotation"]["object"]
    tmp_obj = copy.deepcopy(tmp_obj_list[0])
    tmp_obj["name"] = cls_add
    tmp_obj["bndbox"]["xmin"]=bbox_add[0]
    tmp_obj["bndbox"]["ymin"]=bbox_add[1]
    tmp_obj["bndbox"]["xmax"]=bbox_add[2]
    tmp_obj["bndbox"]["ymax"]=bbox_add[3]
    tmp_obj_list.append(tmp_obj)
 
    xml_dic["annotation"]["object"] = tmp_obj_list
    xmlstr = xmltodict.unparse(xml_dic)
    xml_pretty_str = minidom.parseString(xmlstr).toprettyxml()
    
    with open(xml_path,"w",encoding="utf8") as f:
        f.write(xml_pretty_str)
    
    pbar.update(1)
pbar.close()

3.指定帧区间删除特定的bbox

"""
IOU计算
+ input
  + box1:[box1_x1,box1_y1,box1_x2,box1_y2]
  + box2:[box2_x1,box2_y1,box2_x2,box2_y2]
+ output
  + iou值
"""

def cal_iou(box1,box2):
    # 判断是否能相交
    if abs(box2[2]+box2[0]-box1[2]-box1[0])>box2[2]-box2[0]+box1[2]-box1[0]:
        return 0
    if abs(box2[3]+box2[1]-box1[3]-box1[1])>box2[3]-box2[1]+box1[3]-box1[1]:
        return 0
    
    # 求相交区域左上角的坐标和右下角的坐标
    box_intersect_x1 = max(box1[0], box2[0])
    box_intersect_y1 = max(box1[1], box2[1])
    box_intersect_x2 = min(box1[2], box2[2])
    box_intersect_y2 = min(box1[3], box2[3])
    
    # 求二者相交的面积
    area_intersect = (box_intersect_y2 - box_intersect_y1) * (box_intersect_x2 - box_intersect_x1)
    
    # 求box1,box2的面积
    area_box1 = (box1[2] - box1[0]) * (box1[3] - box1[1])
    area_box2 = (box2[2] - box2[0]) * (box2[3] - box2[1])
    
    # 求二者相并的面积
    area_union = area_box1 + area_box2 - area_intersect
    
    # 计算iou（交并比）
    iou = area_intersect / area_union
    
    return iou
# 删除指定的bbox

import xmltodict
import os
from xml.dom import minidom
from tqdm import tqdm

xml_dir="./data_handle/xml/"# VOC xml文件所在文件夹

# 匹配待删除的bbox
xml = """
    <object>
        <name>Person</name>
        <bndbox>
            <xmin>901</xmin>
            <ymin>6</ymin>
            <xmax>920</xmax>
            <ymax>31</ymax>
        </bndbox>
    </object>
"""

xml_dict = xmltodict.parse(xml)
obj = xml_dict["object"]
cls_filter = obj["name"]
bndbox = obj["bndbox"]
bbox_filter = [int(bndbox["xmin"]),int(bndbox["ymin"]),int(bndbox["xmax"]),int(bndbox["ymax"])]
iou_filter = 0.6


# 起止帧id
frame_begin = 85000
frame_end = 85427

# 加入进度条支持
pbar = tqdm(total=len(os.listdir(xml_dir)))
pbar.set_description("VOC xml filter") # 设置前缀

# 逐一处理xml文件
for file in os.listdir(xml_dir):
    if not file.endswith(".xml"):
        pbar.update(1)
        continue # 过滤掉非xml文件
    
    if int(file[:-4])<frame_begin or int(file[:-4])>frame_end:
        pbar.update(1)
        continue

    xml_path = os.path.join(xml_dir,file) # 拼接xml地址
    
    # 读取xml文件+转为字典
    with open(xml_path,'r',encoding="utf8") as f:
        xml_str = f.read()
    xml_dic = xmltodict.parse(xml_str)
    # 删除掉待过滤的类别
    tmp_obj_list = xml_dic["annotation"]["object"]
    tmp_obj_list_fiter = []
    for tmp_obj in tmp_obj_list:
        tmp_bndbox = tmp_obj["bndbox"]
        tmp_obj_bbox = [int(tmp_bndbox["xmin"]),int(tmp_bndbox["ymin"]),int(tmp_bndbox["xmax"]),int(tmp_bndbox["ymax"])]
        if tmp_obj["name"]==cls_filter and cal_iou(bbox_filter,tmp_obj_bbox)>iou_filter:
            print(cal_iou(bbox_filter,tmp_obj_bbox))
            pass
        else:
            tmp_obj_list_fiter.append(tmp_obj)
    
    xml_dic["annotation"]["object"] = tmp_obj_list_fiter
    xmlstr = xmltodict.unparse(xml_dic)
    xml_pretty_str = minidom.parseString(xmlstr).toprettyxml()
    
    with open(xml_path,"w",encoding="utf8") as f:
        f.write(xml_pretty_str)
    
    pbar.update(1)
pbar.close()

4.指定帧区间修改指定bbox的类别

"""
IOU计算
+ input
  + box1:[box1_x1,box1_y1,box1_x2,box1_y2]
  + box2:[box2_x1,box2_y1,box2_x2,box2_y2]
+ output
  + iou值
"""

def cal_iou(box1,box2):
    # 判断是否能相交
    if abs(box2[2]+box2[0]-box1[2]-box1[0])>box2[2]-box2[0]+box1[2]-box1[0]:
        return 0
    if abs(box2[3]+box2[1]-box1[3]-box1[1])>box2[3]-box2[1]+box1[3]-box1[1]:
        return 0
    
    # 求相交区域左上角的坐标和右下角的坐标
    box_intersect_x1 = max(box1[0], box2[0])
    box_intersect_y1 = max(box1[1], box2[1])
    box_intersect_x2 = min(box1[2], box2[2])
    box_intersect_y2 = min(box1[3], box2[3])
    
    # 求二者相交的面积
    area_intersect = (box_intersect_y2 - box_intersect_y1) * (box_intersect_x2 - box_intersect_x1)
    
    # 求box1,box2的面积
    area_box1 = (box1[2] - box1[0]) * (box1[3] - box1[1])
    area_box2 = (box2[2] - box2[0]) * (box2[3] - box2[1])
    
    # 求二者相并的面积
    area_union = area_box1 + area_box2 - area_intersect
    
    # 计算iou（交并比）
    iou = area_intersect / area_union
    
    return iou

# 修改指定框的类别

import xmltodict
import os
from xml.dom import minidom
from tqdm import tqdm


xml_dir="./data_handle/xml/"# VOC xml文件所在文件夹

# 匹配待修改的bbox

xml = """
    <object>
        <name>RefuelVehicle</name>
        <pose>Unspecified</pose>
        <truncated>0</truncated>
        <difficult>0</difficult>
        <bndbox>
            <xmin>106</xmin>
            <ymin>389</ymin>
            <xmax>237</xmax>
            <ymax>542</ymax>
        </bndbox>
    </object>
"""

xml_dict = xmltodict.parse(xml)
obj = xml_dict["object"]
cls_filter_src = obj["name"]
bndbox = obj["bndbox"]
bbox_filter = [int(bndbox["xmin"]),int(bndbox["ymin"]),int(bndbox["xmax"]),int(bndbox["ymax"])]


cls_filter_dst = "LuggageVehicle"
iou_filter = 0.99

# 起止帧id
frame_begin = 59680
frame_end = 60000

# 加入进度条支持
pbar = tqdm(total=len(os.listdir(xml_dir)))
pbar.set_description("VOC xml filter") # 设置前缀

# 逐一处理xml文件
for file in os.listdir(xml_dir):
    if not file.endswith(".xml"):
        pbar.update(1)
        continue # 过滤掉非xml文件
    
    if int(file[:-4])<frame_begin or int(file[:-4])>frame_end:
        pbar.update(1)
        continue
        
    xml_path = os.path.join(xml_dir,file) # 拼接原始地址
    
    # 读取xml文件+转为字典
    with open(xml_path,'r',encoding="utf8") as f:
        xml_str = f.read()
    xml_dic = xmltodict.parse(xml_str)
    # 删除掉待过滤的类别
    tmp_obj_list = xml_dic["annotation"]["object"]
    tmp_obj_list_fiter = []
    for tmp_obj in tmp_obj_list:
        tmp_bndbox = tmp_obj["bndbox"]
        tmp_obj_bbox = [int(tmp_bndbox["xmin"]),int(tmp_bndbox["ymin"]),int(tmp_bndbox["xmax"]),int(tmp_bndbox["ymax"])]
        if tmp_obj["name"]==cls_filter_src and cal_iou(bbox_filter,tmp_obj_bbox)>iou_filter:
            tmp_obj["name"]=cls_filter_dst
            tmp_obj_list_fiter.append(tmp_obj)
        else:
            tmp_obj_list_fiter.append(tmp_obj)
    
    xml_dic["annotation"]["object"] = tmp_obj_list_fiter
    xmlstr = xmltodict.unparse(xml_dic)
    xml_pretty_str = minidom.parseString(xmlstr).toprettyxml()
    
    with open(xml_path,"w",encoding="utf8") as f:
        f.write(xml_pretty_str)
    
    pbar.update(1)
pbar.close()
0
暂无标签
版权属于： jupiter
本文链接： http://blog.inat.top/archives/369/
作品采用： 《 署名-非商业性使用-相同方式共享 4.0 国际 (CC BY-NC-SA 4.0) 》许可协议授权

评论 (0)

打卡
取消