每天讲解一点PyTorch 【7】np.transpose torch.from_numpy

今天开始讲解：
numpy中transpose支持高维度
np.transpose、torch.from_numpy、.float()

 img = cv2.imread(img_path)
 img.shape
 cv2.imshow(img)
 img = img  / 255.
 img = np.transpose(img, (2, 0, 1)) # numpy中transpose支持高维度
 # numpy转换成tensor
 img = torch.from_numpy(img).float()

再代码测试：

>>> import numpy as np
>>> x=np.random.rand(1,2,3)
>>> 
>>> x
array([[[0.90384323, 0.84000258, 0.07755986],
        [0.59310169, 0.7845629 , 0.0288706 ]]])
>>> x.shape
(1, 2, 3)
>>> 
>>> y = np.transpose(x, (2, 0, 1))
>>> y
array([[[0.90384323, 0.59310169]],

       [[0.84000258, 0.7845629 ]],

       [[0.07755986, 0.0288706 ]]])
>>> 
>>> y.shape
(3, 1, 2)
>>>