mirror of
https://github.com/luzhisheng/js_reverse.git
synced 2025-04-19 09:24:46 +08:00
23 lines
652 B
Python
23 lines
652 B
Python
from torchvision.datasets import MNIST
|
|
from torchvision import transforms
|
|
from torch.utils.data import DataLoader
|
|
from torchvision.utils import make_grid
|
|
import matplotlib.pyplot as plt
|
|
|
|
# 图片标准化处理目的就是提高识别准确度，比如灰度，旋转，等等
|
|
my_transforms = transforms.Compose(
|
|
[
|
|
transforms.PILToTensor(),
|
|
]
|
|
)
|
|
|
|
mnist_train = MNIST(root="./MNIST_data", train=True, download=True, transform=my_transforms)
|
|
|
|
a = DataLoader(mnist_train, batch_size=2, shuffle=True)
|
|
for img, labels in a:
|
|
print(labels)
|
|
image = make_grid(img).permute(1, 2, 0).numpy()
|
|
plt.imshow(image)
|
|
plt.show()
|
|
exit()
|