1. Halcon22.11 download
Download the installation package from the official website (registration is required to download)
Tip: It is jointly developed with VS2019, so you need to install VS2019
Refer to the specific installation steps: https://zhuanlan.zhihu.com/p/590352632
2. CUDA, cudnn download preparation
Relevant installation packages such as CUDA and cudnn versions need to be downloaded and installed locally.
CUDA version selection, take my own as an example: In the NVIDIA control panel, you can see that the RTX3060 driver currently supports up to CUDA 11.7 version
Actual measurement: Currently RTX3060 does not support CUDA version 11 or below; the computing power of cuda should support the computing power of your graphics card: RTX3060 graphics card computing power is 8.6.
The CUDA 11.7 version I chose, download the corresponding computing power cudnn8.6 version
3. CUDA download and install
3.1 CUDA download
Enter CUDA official website to download CUDA, here I downloaded CUDA11.7.1 version
3.2 CUDA installation
Installation reference: CUDA and cuDNN installation tutorial (super detailed) - kylinmin's blog - CSDN blog
Installation failure reference: NVIDIA installer fails, win10 RTX3060 installs CUDA11.7_3060 cuda11.7_zhushun7's blog-CSDN Blog
4.cudnn download and install
4.1cudnn download
Preparations before downloading: You need to check the computing power of the graphics card, and you need to use a tool called deviceQuery.exe, which is under the CUDA directory.
CUDA 11.7 is used locally, and the tool directory is
//根据自己的安装的CUDA版本操作
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\extras\demo_suite>
It can be seen from the above that the computing power of the graphics card is 8.6.
Enter cudnn official website to download cudnn (here you need to register and log in yourself), here I downloaded cudnn v8.6.0 version, select the version corresponding to the CUDA installed above.
4.1 cudnn installation
Refer to the CUDA installation reference above.