Target detection learning process

1. install Anaconda And use

• https://blog.csdn.net/tqlisno1/article/details/108908775?
  conda Basic grasp of ：

2.cuda and cudnn The download 、 install 、 Link and environment variable configuration

• Download and install ：
  https://blog.csdn.net/jhsignal/article/details/111401628
  https://blog.csdn.net/m0_37605642/article/details/98854753
• cudnn Registration failed , Thunder download
  https://blog.csdn.net/qq_40004311/article/details/105076869

3.pytorch install

• https://blog.csdn.net/qq_46092061/article/details/119153893?

 testing pytorch Is it installed well     
import torch      
torch.cuda.is_available()

4. install Vscode

• https://blog.csdn.net/weixin_43883917/article/details/113867914?

Or install pycharm

• https://blog.csdn.net/weixin_46211269/article/details/119934323?
• Download and configure the virtual created before python Environmental Science

 install  jupyter notebook      
conda install nb_canda

5. Study Pytorch

1. python Two magic weapons ： dir() help()
2. Three ways of running code ：python file 、python Console 、jupyter（ Each has its own advantages , A mixture of ）

6. Study YOLOV5
First install the necessary packages , The author provides summary documents or what is missing
（1） forecast detect.py Understanding and setting of parameters
（2） Training train.py Understanding and setting of parameters
（3） Local GPU Training and Cloud GPU Training
Google Colab
（4） Homemade data sets And training
https://www.makesense.ai/
（5） Call the computer camera
https://blog.csdn.net/Tshaxz/article/details/121735482?

7.Nvidia Jetson nano

• Maker blog tutorial ：https://www.yahboom.com/study/jetson-nano

• Linux Common commands ：https://blog.csdn.net/qq_23329167/article/details/83856430?

• First step ： Mirror image burn
  Image file （ Thunder download ）、SD Card Formatter format （ Download from the official website ）、Etcher Write image （ Thunder download ）

• The second step ：windows Lower installation **Xmanager7（xshell、xftp）** Cracked versions
  https://blog.csdn.net/qq_44747572/article/details/121201512?
  （1） problem ：xshell Large delay energy , Maybe the network card is too slow

• The third step ： To configure linux The necessary environment
  （1） Replace the software source https://blog.csdn.net/qq_35451572/article/details/79516563?
  Specify source ： Artifact ：
  https://www.jianshu.com/p/6f9e88e73ca6
  pip install torch torchvision -i http://mirrors.aliyun.com/pypi/simple --trusted-host mirrors.aliyun.com（ Artifact ！！！！！）
  （2）

• Step four ： Deploy yolov5
  linux Command line forced end ：ctrl+z
  （1） The configuration process ：
  https://blog.csdn.net/IamYZD/article/details/119618950?
  （2） see cuda edition ：https://blog.csdn.net/qq_40329272/article/details/106879663
  （3）pip install and sudo apt-get install difference
  https://blog.csdn.net/weixin_38145317/article/details/81168814
  （4）Pip and Pip3 The difference between
  （5） doubt ： use csdn Install the package step by step in the tutorial , And direct use pip download YOLOV5 The author gives .txt What's the difference between files ？
  sudo apt-get install xxx no way , however pip3 install xxx That's ok

https://blog.csdn.net/xiaoyuan2157/article/details/120932987
ubuntu18.04 install ROS：
https://blog.csdn.net/qq_41450811/article/details/99079041?