YOLOv8训练自己的数据集

Charley_3397

于 2024-09-07 21:19:02 发布

阅读量692

点赞数 10

文章标签： YOLO python 机器学习

本文链接：https://blog.csdn.net/qq_45050922/article/details/142001672

版权

1. 配置自己的.yaml文件

当加载这个.yaml文件时，能根据里面的路径找到相应的数据。

# 数据集的根目录
path: D:\edge\YOLO\YOLOv8_Official_Version\ultralytics\cfg\datasets

# 训练数据的目录(.jpg目录)
train: D:\edge\YOLO\YOLOv8_Official_Version\ultralytics\cfg\datasets\images\train

# 验证数据的目录(.jpg目录)
val: D:\edge\YOLO\YOLOv8_Official_Version\ultralytics\cfg\datasets\images\val

# 测试数据的目录(.jpg目录)
test: D:\edge\YOLO\YOLOv8_Official_Version\ultralytics\cfg\datasets\images\test
 
# 类别的数量
nc: 6
 
# class names
names:
  0: ore carrier
  1: passenger ship
  2: container ship
  3: bulk cargo carrier
  4: general cargo ship  
  5: fishing boat

2. 标签数据格式转换 xml格式转换成txt格式

# .xml文件转换成.txt文件

import copy
from lxml.etree import Element, SubElement, tostring, ElementTree
import xml.etree.ElementTree as ET
import pickle
import os
from os import listdir, getcwd
from os.path import join

# 检测目标的类别
classes = ["ore carrier", "passenger ship",
           "container ship", "bulk cargo carrier",
           "general cargo ship", "fishing boat"]

CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))

def convert(size, box):
    dw = 1. / size[0]
    dh = 1. / size[1]
    x = (box[0] + box[1]) / 2.0    # (x_min + x_max) / 2.0
    y = (box[2] + box[3]) / 2.0    # (y_min + y_max) / 2.0
    w = box[1] - box[0]   # x_max - x_min
    h = box[3] - box[2]   # y_max - y_min
    x = x * dw
    w = w * dw
    y = y * dh
    h = h * dh
    return (x, y, w, h)

def convert_annotation(image_id):
    # .xml格式文件的地址
    in_file = open('地址1\%s.xml' % (image_id), encoding='UTF-8')

    # 生成的.txt格式文件的地址
    out_file = open('地址2\%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'):
        cls = obj.find('name').text
        
        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')

# .xml格式文件的地址
xml_path = os.path.join(CURRENT_DIR, '地址1/')

# xml列表
img_xmls = os.listdir(xml_path)
for img_xml in img_xmls:
    label_name = img_xml.split('.')[0]
    print(label_name)
    convert_annotation(label_name)

3. 数据集划分 train val test

# 生成train val teat 数据集文件和标签文件

import os, shutil, random
from tqdm import tqdm

def split_img(img_path, label_path, split_list):
    try:
        # Data是要生成的train，val，test的文件夹  ultralytics\cfg\datasets
        Data = '地址1'

        train_img_dir = Data + '/images/train'
        val_img_dir = Data + '/images/val'
        test_img_dir = Data + '/images/test'

        train_label_dir = Data + '/labels/train'
        val_label_dir = Data + '/labels/val'
        test_label_dir = Data + '/labels/test'

        # 创建文件夹
        os.makedirs(train_img_dir)
        os.makedirs(train_label_dir)
        os.makedirs(val_img_dir)
        os.makedirs(val_label_dir)
        os.makedirs(test_img_dir)
        os.makedirs(test_label_dir)

    except:
        print('文件目录已存在')

    train, val, test = split_list
    all_img = os.listdir(img_path)
    all_img_path = [os.path.join(img_path, img) for img in all_img]
    train_img = random.sample(all_img_path, int(train * len(all_img_path)))
    train_img_copy = [os.path.join(train_img_dir, img.split('\\')[-1]) for img in train_img]
    train_label = [toLabelPath(img, label_path) for img in train_img]
    train_label_copy = [os.path.join(train_label_dir, label.split('\\')[-1]) for label in train_label]
    
    for i in tqdm(range(len(train_img)), desc='train ', ncols=80, unit='img'):
        _copy(train_img[i], train_img_dir)
        _copy(train_label[i], train_label_dir)
        all_img_path.remove(train_img[i])
    val_img = random.sample(all_img_path, int(val / (val + test) * len(all_img_path)))
    val_label = [toLabelPath(img, label_path) for img in val_img]
    
    for i in tqdm(range(len(val_img)), desc='val ', ncols=80, unit='img'):
        _copy(val_img[i], val_img_dir)
        _copy(val_label[i], val_label_dir)
        all_img_path.remove(val_img[i])
    test_img = all_img_path
    test_label = [toLabelPath(img, label_path) for img in test_img]
    
    for i in tqdm(range(len(test_img)), desc='test ', ncols=80, unit='img'):
        _copy(test_img[i], test_img_dir)
        _copy(test_label[i], test_label_dir)


def _copy(from_path, to_path):
    shutil.copy(from_path, to_path)


def toLabelPath(img_path, label_path):
    img = img_path.split('\\')[-1]
    label = img.split('.jpg')[0] + '.txt'
    return os.path.join(label_path, label)


if __name__ == '__main__':
    # 输入的图像路径
    img_path = '地址2'
    
    # 输入的标签路径
    label_path = '地址3'

    # 数据集的划分比例
    split_list = [0.81, 0.09, 0.1]
    split_img(img_path, label_path, split_list)

4. 数据集存放结构

路径：ultralytics\cfg\datasets

images(文件夹)

train(文件夹)

val(文件夹)

test(文件夹)

labels(文件夹)

train(文件夹)

val(文件夹)

test(文件夹)

标签数据格式转换.py

数据集划分.py

seaships.yaml(数据加载文件，自定义数据集中所有目标对象的类别信息)

5. 训练数据

路径：ultralytics\cfg\models\v8\yolov8.yaml 模型加载文件只需要修改类别数nc即可

路径：ultralytics 存放一个预训练权重文件 yolov8s.pt

注释：在设置模型规模(n，s，m，l，x)的时候，直接通过Model_yaml参数来设置：model_yaml=r"D:\edge\YOLO\YOLOv8_Official_Version\ultralytics\cfg\models\v8\yolov8s.yaml"，虽然该路径下没有yolov8s.yaml的文件，但是v8可以识别出来选择的模型类型。

注释：可以适当地调小batch和workers，不然可能会报错。

注释：路径：ultralytics\cfg\default.yaml 其他参数可以在这里进行设置

from ultralytics import YOLO

if __name__ == "__main__":
    
    # 加载模型
    model_yaml = r"地址1\yolov8s.yaml"
    data_yaml = r"地址2\seaships.yaml"
    pre_model = r"地址3\yolov8s.pt"

    # build from YAML and transfer weights
    model = YOLO(model_yaml, task='detect').load(pre_model)

    # 训练模型
    results = model.train(data=data_yaml, epochs=100, imgsz=640, batch=4, workers=2)