将本地图片转换成tensor

注意：

• 图片格式：（w，h）或（长 * 宽）
• pytorch支持的tensor格式：（c，h，w）

1. 先将图片转换成ndarray，然后再转换成tensor。

from PIL import  Image
import numpy as np
import torch

image_dir = '0001.png'
img = Image.open(image_dir)
img.show()

img.size  # (w, h)

type(img)

img =np.array(img)  # (w, h) ---> (h, w, c)
img.shape

img = img.transpose(2, 0, 1)  # (h, w, c) ---> (c, h, w)
img.shape

img = torch.tensor(img)
img.shape

2. 使用torchvision，做图像增广。

from PIL import Image
import torch
from torchvision import transforms

image_dir = '0001.png'
img = Image.open(image_dir)
# img.show()

img.size  # (w, h)

transforms.ToTensor():

• Converts a PIL Image or numpy.ndarray (H x W x C) in the range
  [0, 255] to a torch.FloatTensor of shape (C x H x W) in the range [0.0, 1.0]

img_transform = transforms.Compose([transforms.ToTensor()])
img = img_transform(img)

img.shape