学习pytorch 2
2. dataset实战
B站小土堆视频
代码
py
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()