yolov8 数据集划分代码

### YOLOv8 数据集划分

为了实现YOLOv8的数据集划分，通常会采用Python脚本处理标签文件和图像路径。下面提供一段用于将VOC格式转换为YOLO格式并进行训练/验证集划分的代码示例：

import os
from xml.etree import ElementTree as ET
import random
import shutil

def convert_voc_to_yolo(voc_annotation_file, output_dir, class_names):
    tree = ET.parse(voc_annotation_file)
    root = tree.getroot()
    
    image_name = root.find('filename').text
    
    width = int(root.find('size/width').text)
    height = int(root.find('size/height').text)

    yolo_lines = []
    for obj in root.findall('object'):
        label = obj.find('name').text
        
        if label not in class_names:
            continue
            
        index = class_names.index(label)
        
        bbox = obj.find('bndbox')
        xmin = float(bbox.find('xmin').text)
        ymin = float(bbox.find('ymin').text)
        xmax = float(bbox.find('xmax').text)
        ymax = float(bbox.find('ymax').text)
        
        x_center = (xmin + xmax) / 2.0 / width
        y_center = (ymin + ymax) / 2.0 / height
        w_bbox = (xmax - xmin) / width
        h_bbox = (ymax - ymin) / height
        
        line = f"{index} {x_center:.6f} {y_center:.6f} {w_bbox:.6f} {h_bbox:.6f}\n"
        yolo_lines.append(line)
    
    with open(os.path.join(output_dir, os.path.splitext(image_name)[0] + '.txt'), 'w') as file:
        file.writelines(yolo_lines)


def split_dataset(input_folder, output_train_folder, output_val_folder, train_ratio=0.8):
    all_files = [os.path.join(input_folder, f) for f in os.listdir(input_folder)]
    images = sorted([file for file in all_files if file.lower().endswith('.jpg')])
    annotations = sorted([file.replace('.jpg', '.xml') for file in images])

    combined = list(zip(images, annotations))
    random.shuffle(combined)
    images[:], annotations[:] = zip(*combined)

    num_images = len(images)
    split_index = int(num_images * train_ratio)

    # 创建目标目录
    os.makedirs(output_train_folder, exist_ok=True)
    os.makedirs(output_val_folder, exist_ok=True)

    for i, (image_path, annotation_path) in enumerate(zip(images, annotations)):
        destination_folder = output_train_folder if i < split_index else output_val_folder
        shutil.copy(image_path, destination_folder)
        convert_voc_to_yolo(annotation_path, destination_folder, ["class1", "class2"])  # 替换类别名称列表


if __name__ == "__main__":
    input_directory = "/path/to/voc_format_data"  # 输入VOC格式数据的位置
    training_output = "./data/train/"
    validation_output = "./data/validation/"

    split_dataset(input_directory, training_output, validation_output)

此段代码实现了两个主要功能：一是`convert_voc_to_yolo()`函数负责读取XML文件并将边界框坐标从Pascal VOC格式转换成YOLO所需的相对位置表示；二是`split_dataset()`函数按照指定比例随机分配图片到训练集和验证集中。