pytorch中Tensor和ToTensor

import torch
from torchvision import transforms
__image_transform = transforms.Compose([
            transforms.ToTensor()])
img=np.array(Image.open(img_path)) / 255. - 0.5#打开图片
img_data = torch.Tensor(img)  #转成Tensor

img_data = __image_transform(img)  #转成ToTensor       

打开的图片img是w,h,c形状的三维numpy数组
1.
img_data = torch.Tensor(img) #转成Tensor
得到的结果是：w,h,c形状的三维tensor
2.
img_data = __image_transform(img) #转成ToTensor
得到的结果是：c,h,w形状的三维tensor