清华源安装cpu版本得pytorch：conda install pytorch torchvision cpuonly pytorch
opencv的像素是BGR顺序，然而matplotlib所遵循的是RGB顺序，pytorch BGR image
https://www.cnblogs.com/visionfeng/p/6094423.html
img/255 进行图像归一化，值阈为[0, 1]
https://blog.csdn.net/qq_45704645/article/details/111088728?utm_medium=distribute.pc_relevant_t0.none-task-blog-2%7Edefault%7EBlogCommendFromMachineLearnPai2%7Edefault-1.control&depth_1-utm_source=distribute.pc_relevant_t0.none-task-blog-2%7Edefault%7EBlogCommendFromMachineLearnPai2%7Edefault-1.control
tensor数据与numpy转换
https://blog.csdn.net/cunchi4221/article/details/107471992?utm_medium=distribute.pc_relevant.none-task-blog-2%7Edefault%7EBlogCommendFromMachineLearnPai2%7Edefault-9.control&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2%7Edefault%7EBlogCommendFromMachineLearnPai2%7Edefault-9.control
a.permute(2,1,0) # 按照第2、 1、 0的维度顺序重新进行元素排列
安装netron pip install netron ，查看安装位置 pip install netron
使用netron
import netron
modelPath = "你的模型文件名.扩展名"
netron.start(modelPath)
Batchnorm 归一化实现了数据之间绝对差异的最小化，相对差异的扩大化
https://blog.csdn.net/qq_25737169/article/details/79048516
ubuntu网络配置
https://blog.csdn.net/mary19920410/article/details/71191954/?utm_medium=distribute.pc_relevant.none-task-blog-baidujs_baidulandingword-1&spm=1001.2101.3001.4242
