看代码吧~

import torch
import numpy as np
from torchvision.transforms import ToTensor 
t = torch.tensor(np.arange(24).reshape(2,4,3))
print(t)
 
#HWC 转CHW
print(t.transpose(0,2).transpose(1,2)) 
print(t.permute(2,0,1)) 
print(ToTensor()(t.numpy()))

D:\anaconda\python.exe C:/Users/liuxinyu/Desktop/pytorch_test/day3/hwc转chw.py
tensor([[[ 0, 1, 2],
[ 3, 4, 5],
[ 6, 7, 8],
[ 9, 10, 11]],

[[12, 13, 14],
[15, 16, 17],
[18, 19, 20],
[21, 22, 23]]], dtype=torch.int32)
tensor([[[ 0, 3, 6, 9],
[12, 15, 18, 21]],

[[ 1, 4, 7, 10],
[13, 16, 19, 22]],

[[ 2, 5, 8, 11],
[14, 17, 20, 23]]], dtype=torch.int32)
tensor([[[ 0, 3, 6, 9],
[12, 15, 18, 21]],

[[ 1, 4, 7, 10],
[13, 16, 19, 22]],

[[ 2, 5, 8, 11],
[14, 17, 20, 23]]], dtype=torch.int32)
tensor([[[ 0, 3, 6, 9],
[12, 15, 18, 21]],

[[ 1, 4, 7, 10],
[13, 16, 19, 22]],

[[ 2, 5, 8, 11],
[14, 17, 20, 23]]], dtype=torch.int32)

Process finished with exit code 0

补充：opencv python 把图(cv2下)BGR转RGB，且HWC转CHW

如下所示：

img = cv2.imread("001.jpg")
img_ = img[:,:,::-1].transpose((2,0,1))

① 在opencv里，图格式HWC，其余都是CHW，故transpose((2,0,1))

② img[:,:,::-1]对应H、W、C，彩图是3通道，即C是3层。opencv里对应BGR，故通过C通道的 ::-1 就是把BGR转为RGB

注： [::-1] 代表顺序相反操作

③ 若不涉及C通道的BGR转RGB，如Img[:,:,0]代表B通道，也就是蓝色分量图像；Img[:,:,1]代表G通道，也就是绿色分量图像；

Img[:,:,2]代表R通道，也就是红色分量图像。

补充：python opencv 中将图像由BGR转换为CHW用于后期的深度训练

BGR HWC -> CHW 12 -> HCW 01 -> CHW

import cv2 as cv
import  numpy as np
img = cv.imread("lenna.png")
#BGR HWC -> CHW  12 -> HCW 01 -> CHW
transform_img = img.swapaxes(1,2).swapaxes(0,1)
print(img.shape)
print(transform_img.shape)
cv.imshow("image0 ",transform_img[0])
cv.imshow("image1",transform_img[1])
cv.imshow("image2",transform_img[2])
cv.waitKey(0)
cv.destroyAllWindows()

以上为个人经验，希望能给大家一个参考，也希望大家多多支持hwidc。