pytorch將自己數據集用去訓練，dataloader

from torch.utils.data import DataLoader, Dataset
from torchvision import transforms
import pandas as pd
import cv2
import os
import glob


class Mydataset(Dataset):
    def __init__(self, data_dir, cls_list, transform=None, suffix="*.jpg"):
        super().__init__()
        self.data_dir = data_dir
        file_paths = []
        labels = []
        for index, cls_name in enumerate(cls_list):
            file_list = glob.glob(os.path.join(data_dir, cls_name, suffix))
            if file_list:
                file_paths.extend(file_list)
                labels.extend([index for i in file_list])
        self.df = pd.DataFrame({
            "file_paths": file_paths,
            "labels": labels
        },
                               dtype='object').values
        self.transform = transform

    def __len__(self):
        return len(self.df)

    def __getitem__(self, idex):
        img_name, label = self.df[idex]
        image = cv2.imread(img_name)
        if image.shape[2] == 1:
            image = cv2.cvtColor(image, cv2.COLOR_GRAY2RGB)
        if self.transform is not None:
            image = self.transform(image)
        return image, label


transforms_train = transforms.Compose([
    transforms.ToPILImage(),
    transforms.Resize((256, 256)),
    transforms.RandomHorizontalFlip(),  # 水平翻轉
    transforms.RandomRotation(10),  # 隨機旋轉10度
    transforms.ToTensor(),  # 將數據轉換成Tensor型
    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])

if __name__ == "__main__":
    train_path = '/Users/goby/data/tianshi_image/train_img'
    train_data = Mydataset(train_path, ["金", "木", "水", "火", "土"], transform=transforms_train)
    BATCH_SIZE = 64
    dataloader = DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=False)
    for i_batch, sample_batched in enumerate(dataloader, 0):
        print(i_batch)