数据集清理工具，清理无对应label的图片（也就是没有标注的）

import os


def check_and_clean(image_dir, label_dir):
    missing_labels = []
    for root, _, files in os.walk(image_dir):
        for file in files:
            if file.endswith('.png'):  # 根据你的图片格式调整
                image_path = os.path.join(root, file)

                # 生成标签文件路径
                file_name_without_ext = os.path.splitext(file)[0]
                label_path = os.path.join(label_dir, f"{file_name_without_ext}.txt")

                if not os.path.exists(label_path):
                    missing_labels.append(file)
                    # 删除没有对应标签的图片
                    os.remove(image_path)
                    print(f"Deleted {image_path}")

    return missing_labels


# 设置图片和标签目录
image_dir = r'C:\Users\合格的牛马\Desktop\datasets\wangzherongyao\images\val'
label_dir = r'C:\Users\合格的牛马\Desktop\datasets\wangzherongyao\labels\val'

# 检查标签文件并清理图片
missing_labels = check_and_clean(image_dir, label_dir)

# 输出缺少标签的图像文件
if missing_labels:
    print('Missing labels for the following images:')
    for image in missing_labels:
        print(image)
else:
    print('All images have corresponding label files.')