PILImage 和 Tensor转换 TypeError: Input type int64 is not supported

# 报错代码

import torch
import torchvision
from PIL import Image

img = Image.open('./post_disaster.png')

tensor = torchvision.transforms.ToTensor()(img)      # 转换为Tensor

tensor = tensor.to(torch.long)

img = torchvision.transforms.ToPILImage()(tensor)  

img.show()

出错原因: transform.ToPILImage()(tensor) 该方法接受的tensor需要是torch.float32类型的参数, 不能是torch.int类型的参数, 修改tensor的dtype类型为torch.float32, 否则会报错如下图