HMDB51数据集划分训练集测试集

最新推荐文章于 2024-07-12 16:55:56 发布

Will_u_win

最新推荐文章于 2024-07-12 16:55:56 发布

阅读量428

点赞数

文章标签：人工智能 python 机器学习

本文链接：https://blog.csdn.net/Will_u_win/article/details/134935389

版权

import os
import shutil
import random

def split_dataset(dataset_path, train_ratio=0.8, val_ratio=0.1, test_ratio=0.1):
    class_folders = sorted(os.listdir(dataset_path))

    for class_folder in class_folders:
        class_path = os.path.join(dataset_path, class_folder)
        video_folders = sorted(os.listdir(class_path))

        num_videos = len(video_folders)
        num_train = int(num_videos * train_ratio)
        num_val = int(num_videos * val_ratio)
        num_test = num_videos - num_train - num_val

        random.shuffle(video_folders)

        train_videos = video_folders[:num_train]
        val_videos = video_folders[num_train:num_train + num_val]
        test_videos = video_folders[num_train + num_val:]

        for video_folder in train_videos:
            src_path = os.path.join(class_path, video_folder)
            dest_path = os.path.join("train", class_folder, video_folder)
            shutil.move(src_path, dest_path)

        for video_folder in val_videos:
            src_path = os.path.join(class_path, video_folder)
            dest_path = os.path.join("val", class_folder, video_folder)
            shutil.move(src_path, dest_path)

        for video_folder in test_videos:
            src_path = os.path.join(class_path, video_folder)
            dest_path = os.path.join("test", class_folder, video_folder)
            shutil.move(src_path, dest_path)

if __name__ == "__main__":
    dataset_path = "/home/db/下载/HMDB51"
    train_ratio = 0.8
    val_ratio = 0.1
    test_ratio = 0.1

    # Create train, val, and test directories
    os.makedirs("train", exist_ok=True)
    os.makedirs("val", exist_ok=True)
    os.makedirs("test", exist_ok=True)

    split_dataset(dataset_path, train_ratio, val_ratio, test_ratio)