# Compile CUDA on OpenCV

**URL:** <https://forum.opencv.org/t/compile-cuda-on-opencv/18617>\
**Category:** C++\
**Tags:** dnn, windows, build, cuda\
**Created:** [August 30, 2024, 2:04am UTC](https://forum.opencv.org/t/compile-cuda-on-opencv/18617 "2024-08-30T02:04:58Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![t2chiang](https://avatars.discourse-cdn.com/v4/letter/t/5e9695/32.png) [@t2chiang](https://forum.opencv.org/u/t2chiang)\
**Post date:** [August 30, 2024, 2:04am UTC](https://forum.opencv.org/t/compile-cuda-on-opencv/18617/1 "2024-08-30T02:04:58Z")

</div>

Dear all, I want to ask some help to build the CUDA version OpenCV in my device. Here are my environment and device details:

GPU: NVIDIA GetForce RTX 3060  
CUDA: 12.1  
Cudnn: 8.9.7  
CMAKE: 3.26.3  
Opencv: 4.10.0  
Generator: Visual Studio 17 2022

When compiling, I selected both WITH\_CUDA and OPENCV\_DNN\_CUDA, and then CUDA\_ARC\_BIN is set up as 8.6. After compiling, there is no error in CMAKE. But there is error when building on Visual Studio. It always say windows SDK 10.0.22 was not found. But when I check the installer of Visual Studio, I cannot find this SDK to install. Does anyone have relevent experience on this issue? Thank you guys