学习pytorch2 导入查看dataset

2. dataset实战

B站小土堆视频

代码

from torch.utils.data import Dataset
from PIL import Image
#import cv2
import os

class MyData(Dataset):

    def __init__(self, root_dir, label_dir):
        self.root_dir = root_dir
        self.label_dir = label_dir
        self.path = os.path.join(self.root_dir, self.label_dir)
        self.img_list = os.listdir(self.path)

    def __getitem__(self, idx):
        img_name = self.img_list[idx]
        img_path = os.path.join(self.root_dir, self.label_dir, img_name)
        img = Image.open(img_path)
        label = self.label_dir
        return img, label

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

if __name__ == '__main__':
    root_dir = "hymenoptera_data/train"
    ants_label_dir = "ants"
    bees_label_dir = "bees"
    ants_dataset = MyData(root_dir, ants_label_dir)
    bees_dataset = MyData(root_dir, bees_label_dir)
    train_dataset = ants_dataset + bees_dataset
    img, label = ants_dataset[1]
    img, label = train_dataset[1]
    img.show()

数据集

https://download.pytorch.org/tutorial/hymenoptera_data.zip