YOLOv5训练自己的数据集

墨蓝 2023-01-06 05:45 242阅读 0赞

## 一、准备工作 ##

在data下新建几个文件夹  
![在这里插入图片描述][20210114092311280.png]  
XML文件放到Annotations  
图片文件放到images

## 二、xml转txt（xml格式的label转换为yolo版的label） ##

# -*- coding:utf-8 -*-
    import sys
    sys.path.append('D:\\Program files\\Anaconda\\libs')
    import os #os：操作系统相关的信息模块
    import random #导入随机函数
    #存放原始图片地址
    data_base_dir = "./data/images"
    file_list = [] #建立列表，用于保存图片信息
    #读取图片文件，并将图片地址、图片名和标签写到txt文件中
    write_file_name = 'ImageXML.txt'
    write_file = open(write_file_name, "w") #以只写方式打开write_file_name文件
    for file in os.listdir(data_base_dir): #file为current_dir当前目录下图片名
        if file.endswith(".jpg"): #如果file以jpg结尾
            write_name = file #图片路径 + 图片名 + 标签
            file_list.append(write_name) #将write_name添加到file_list列表最后
        sorted(file_list) #将列表中所有元素随机排列
        number_of_lines = len(file_list) #列表中元素个数
        #将图片信息写入txt文件中，逐行写入
    for current_line in range(number_of_lines):
        write_file.write(file_list[current_line][:-4] + '\n')#关闭文件

生成ImageXML.txt文件，格式是下面这样的（即图片名称）：  
![在这里插入图片描述][2021011409271569.png]

## 三、运行voc\_label.py ##

# 坐标xml转txt
    import os
    import xml.etree.ElementTree as ET
    
    classes = ["040500000","040500013","040500021","040500022","040500023","040500031","040500032","040500033"]  # 输入名称，必须与xml标注名称一致
    
    def convert(size, box):
        print(size, box)
        dw = 1. / size[0]
        dh = 1. / size[1]
        x = (box[0] + box[1]) / 2.0
        y = (box[2] + box[3]) / 2.0
        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)
    
    
    def convert_annotation(image_id):
        print(image_id)
        in_file = open(r'./data/Annotations/%s.xml' % (image_id), 'rb')  # 读取xml文件路径
    
        out_file = open('./data/labels/%s.txt' % (image_id), 'w')  # 需要保存的txt格式文件路径
        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'):
            cls = obj.find('code').text
            if cls not in classes:  # 检索xml中的名称
                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')
    
    
    image_ids_train = open('./ImageXML.txt').read().strip().split()  # 读取xml文件名索引
    
    for image_id in image_ids_train:
        print(image_id)
        convert_annotation(image_id)

需要注意，若是有中文路径的话，请这样读文件：

open(r'./data/Annotations/%s.xml' % (image_id), 'rb')

接下来就会在data/labels中看到：所有的txt标签  
![在这里插入图片描述][20210114093222414.png]

## 四、切分训练集与测试集 ##

执行train\_test\_split.py

import os
    import random
    
    trainval_percent = 0.1  # 可自行进行调节
    train_percent = 0.9
    xmlfilepath = './data/Annotations'
    txtsavepath = './data/ImageSets/Main'
    total_xml = os.listdir(xmlfilepath)
    
    num = len(total_xml)
    list = range(num)
    tv = int(num * trainval_percent)
    tr = int(tv * train_percent)
    trainval = random.sample(list, tv)
    train = random.sample(trainval, tr)
    
    # ftrainval = open('ImageSets/Main/trainval.txt', 'w')
    ftrain = open('./data/ImageSets/Main/train.txt', 'w')
    ftest = open('./data/ImageSets/Main/test.txt', 'w')
    
    # fval = open('ImageSets/Main/val.txt', 'w')
    
    for i in list:
        name = total_xml[i][:-4] + '\n'
        if i in trainval:
            # ftrainval.write(name)
            if i in train:
                ftest.write(name)
            # else:
            # fval.write(name)
        else:
            ftrain.write(name)
    
    # ftrainval.close()
    ftrain.close()
    # fval.close()
    ftest.close()

## 五、生成训练和测试的txt文件（内容为图片的绝对路径） ##

运行path\_trans，data目录下生成train.txt和test.txt

import xml.etree.ElementTree as ET
    import pickle
    import os
    from os import listdir, getcwd
    from os.path import join
    
    sets = ['train', 'test']
    
    classes = ["040500000","040500013","040500021","040500022","040500023","040500031","040500032","040500033"]  # 输入名称，必须与xml标注名称一致
      # 自己训练的类别
    
    
    def convert(size, box):
        dw = 1. / size[0]
        dh = 1. / size[1]
        x = (box[0] + box[1]) / 2.0
        y = (box[2] + box[3]) / 2.0
        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)
    
    
    def convert_annotation(image_id):
        in_file = open('data/Annotations/%s.xml' % (image_id), 'rb')
        out_file = open('data/labels/%s.txt' % (image_id), 'w')
        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('code').text
            if cls not in classes or int(difficult) == 1:
                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')
    
    
    wd = getcwd()
    for image_set in sets:
        if not os.path.exists('data/labels/'):
            os.makedirs('data/labels/')
        image_ids = open('data/ImageSets/Main/%s.txt' % (image_set)).read().strip().split()
        list_file = open('data/%s.txt' % (image_set), 'w')
        for image_id in image_ids:
            list_file.write('data/images/%s.jpg\n' % (image_id))
            convert_annotation(image_id)
        list_file.close()

![在这里插入图片描述][20210114093845268.png]  
![在这里插入图片描述][20210114095105374.png]

## 六、data目录下新建mydatasets.yaml ##

# train and val datasets (image directory or *.txt file with image paths)
    train: ./data/images  # 上面我们生成的train，根据自己的路径进行更改
    test: ./data/images  # 上面我们生成的test
    #test: ../coco/test-dev2017.txt  # 20k images for submission to https://competitions.codalab.org/competitions/20794
    
    # number of classes
    nc: 8   #训练的类别
    
    # class names
    names : ["040500000","040500013","040500021","040500022","040500023","040500031","040500032","040500033"]

## 七、训练 ##

运行train.py

python train.py --data data/mydatasets.yaml --cfg models/yolov5s.yaml --weights weights/yolov5s.pt --batch-size 16 --epochs 300

[20210114092311280.png]: /images/20221119/8b494b37a50840018afbf4a6dd021a4c.png
[2021011409271569.png]: /images/20221119/a32b4a8d4b8642ebb57d496db60525a6.png
[20210114093222414.png]: /images/20221119/e442456703194cd0b217f89181b09c33.png
[20210114093845268.png]: /images/20221119/9b79b8568c6b492994aaca0775c98ecf.png
[20210114095105374.png]: /images/20221119/3823530128cb4da0afe2421611de4369.png