## 1. Install CUDA and CUDNN
check your CUDA and CUDNN version compatibility of your personal computer
Suggest: CUDA >= 10.0, CUDNN >= 7.5
建議上網找跟自己相同或是差不多電腦規格(CPU, GPU)的人安裝成功的對應版本
Reference: https://medium.com/ching-i/win10-%E5%AE%89%E8%A3%9D-cuda-cudnn-%E6%95%99%E5%AD%B8-c617b3b76deb
## 2. Download Visual Studio
version:
Win11 SDK for Visual Studio 2022(17)
Win10 SDK for Visual Studio 2019(16)
確保有安裝Visual Studio Installer到本地端
Install "Desktop development with C++"
如果要下載舊版(2019)需加入微軟的一個計劃才能免費安裝
Reference: https://visualstudio.microsoft.com/zh-hant/downloads/
## 3. Download Cmake
由於GitHub上的OpenCV Sources只有最基本的函式庫跟功能，因此需要透過Cmake自己build需要的OpenCV函式庫
Reference: https://cmake.org/download/
## 4. Download OpenCV Sources and OpenCV Contrib
download version: 4.7.0
OpenCV Sources: https://opencv.org/releases/
OpenCV Contrib: https://github.com/opencv/opencv_contrib
兩個下載並解壓縮完後，新增一個空的資料夾命名為build，將三者放在同一個路徑下
## 5. Run Cmake
- Source code path and where to build
- select VS and generator(x64)
- click WITH_CUDA, OPENCV_DNN_CUDA, ENABLE_FAST_MATH
- select module folder in OpenCV-Contrib path on OPENCV_EXTRA_MODULES_PATH
- configure again, click CUDA_FAST_MATH
- check your compute capability with your GPU
- select CUDA_ARCH_BIN and enter your version
- configure again, then generate
## 6. Build OpenCV with CUDA
- search build folder
- use Visual Studio open "OpenCV.sln"
- change debug mode to release mode
- click build on ALL_BUILD in CMake Targets
- click build on INSTALL in CMake Targets