Pytorch学习笔记--Transforms的使用（totensor，normalize）

先导入相关库：

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from  torchvision import transforms
from py.Transforms import write

再导入图片：

img_path="D:/Pychram_Project/Learn_Pytorch/Projects/dataset/train/ants/69639610_95e0de17aa.jpg"
img=Image.open(img_path)
#print(img)

totensor的使用，之前已经提过这里不再赘述。

#ToTensor的使用
trans_totensor = transforms.ToTensor() #创建对象
img_tensor=trans_totensor(img)   #转换成imgtensor类型
#print(type(img_tensor))
write.add_image("Totensor",img_tensor)

 Normalize的使用，归一化：注释说明如下

"""Normalize a tensor image with mean and standard deviation.
   This transform does not support PIL Image.
   Given mean: ``(mean[1],...,mean[n])`` and std: ``(std[1],..,std[n])`` for ``n``
   channels, this transform will normalize each channel of the input
   ``torch.*Tensor`` i.e.,
   ``output[channel] = (input[channel] - mean[channel]) / std[channel]``

   .. note::
       This transform acts out of place, i.e., it does not mutate the input tensor.

   Args:
       mean (sequence): Sequence of means for each channel.
       std (sequence): Sequence of standard deviations for each channel.
       inplace(bool,optional): Bool to make this operation in-place.

   """

公式为：output[channel] = (input[channel] - mean[channel]) / std[channel]

print(img_tensor[0][0][0])   #归一化之前的
trans_norm=transforms.Normalize([0.5,0.5,0.5],[0.5,0.5,0.5])
#公式：output[channel] = (input[channel] - mean[channel]) / std[channel]
#在这里（input-0.5）/0.5=2*input-1
img_norm=trans_norm(img_tensor)   #参数只支持image类型
print(img_norm[0][0][0])   #归一化之后的
write.add_image("nomralize",img_norm)
write.close()

在这里传入的根据参数，output=（input-0.5）/0.5=2*input-1   在上文这里归一化之前print输出为tensor(0.9294)，归一化之后输出为tensor(0.8588)，正好符合公式。

归一化这里参数只支持image类型。

查看归一化前后图片对比：

常见的transform：

 

学习内容来自b站小土堆 ，这里只作为学习笔记。