目标检测标签XML转YOLO格式

湫兮之风

已于 2023-12-05 14:32:26 修改

阅读量733

点赞数 7

CC 4.0 BY-SA版权

分类专栏：目标检测文章标签：目标检测 xml YOLO

于 2023-11-23 15:22:10 首次发布

本文链接：https://blog.youkuaiyun.com/Ethan_Rich/article/details/134577934

目标检测专栏收录该内容

2 篇文章

订阅专栏

本文介绍了如何将XML格式的目标检测标注转换为YOLO所需的格式，包括XML标签结构解析、坐标归一化和标签文件生成的Python代码示例。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

XML格式标签

<annotation>
	<folder>images</folder>
	<filename>0001.jpg</filename>
	<path>/Users/Ralf/Downloads/smoking/images/0001.jpg</path>
	<source>
		<database>Unknown</database>
	</source>
	<size>
		<width>460</width>
		<height>480</height>
		<depth>1</depth>
	</size>
	<segmented>0</segmented>
	<object>
		<name>smoking</name>
		<pose>Unspecified</pose>
		<truncated>0</truncated>
		<difficult>0</difficult>
		<bndbox>
			<xmin>156</xmin>
			<ymin>112</ymin>
			<xmax>367</xmax>
			<ymax>333</ymax>
		</bndbox>
	</object>
</annotation>

XML格式大家应该很熟悉，无论是学后端的，还是算法，XML是一种数据的配置格式或者数据的存储格式。对于目标检测来说，我们所需的信息是XML中的<name>标签，它规定了目标检测中的类别索引。<xmin>、<ymin>、<xmax>、<ymax>分别代表左上角右上角的坐标。

YOLO格式标签

0 0.5684782608695652 0.4635416666666667 0.45869565217391306 0.46041666666666664

五个值分别代表类别索引、归一化的中心点x坐标、归一化的中心点y坐标、归一化的框宽度，归一化的框高度。所以想要快速训练yolo，就需要把xml格式的标签改为yolo格式的标签，当然也有的代码中包含标签转换的代码，视情况而定。

转换代码

import xml.etree.ElementTree as ET
import pickle
import os
from os import listdir, getcwd
from os.path import join
import random
from shutil import copyfile

# 根据自己的需求设置类文件
classes = ['类别1name','类别2name']

# Training ratio
# 根据自己的需求设置训练比
# TRAIN_RATIO = 70


# Traversing folders
def clear_hidden_files(path):
    dir_list = os.listdir(path)
    for i in dir_list:
        abspath = os.path.join(os.path.abspath(path), i)
        if os.path.isfile(abspath):
            if i.startswith("._"):
                os.remove(abspath)
        else:
            clear_hidden_files(abspath)


# Normalization of width and height is performed
# Size is the width and height of the original image
def convert(size, box):
    dw = 1. / size[0]
    dh = 1. / size[1]
    # Get Center Point
    x = (box[0] + box[1]) / 2.0
    y = (box[2] + box[3]) / 2.0
    # Calculate width and height
    w = box[1] - box[0]
    h = box[3] - box[2]
    x = x * dw
    w = w * dw
    y = y * dh
    h = h * dh
    return (x, y, w, h)


# Parsing xml files
def convert_annotation(image_id):
    in_file = open('C:/Users/Desktop/Annotations/%s.xml' % image_id, 'rb')
    out_file = open('C:/Users/Desktop/labels/%s.txt' % image_id, 'a')
    tree = ET.parse(in_file)
    root = tree.getroot()
    size = root.find('size')
    w = int(size.find('width').text)
    h = int(size.find('height').text)

    for obj in root.iter('object'):
        difficult = obj.find('difficult').text
        cls = obj.find('name').text
        # if cls not in classes or int(difficult) == 1:
        if cls not in classes:
            continue
        cls_id = classes.index(cls)
        xmlbox = obj.find('bndbox')
        b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text),
             float(xmlbox.find('ymax').text))
        bb = convert((w, h), b)
        out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
    in_file.close()
    out_file.close()


if __name__ == '__main__':
    base_path = r"C:\Users\Desktop\数据集"
    xml_list = os.listdir(base_path)
    for item in xml_list:
        imageid=item.split('.')[0]
        # print(imageid)
        convert_annotation(imageid)