Cmake with_cuda on, and getCudaEnabledDeviceCount () & gt; 0
Vesion:
Cuda 9.1
Opencv3.4
Geforce 940 m support cuda
Description: compile successfully without error but stiching routines in the sample run in seam_finder - & gt; Find (images_warped_f, corners, masks_warped); Function calls when throws exceptions, and successfully solves the bosses have ever had,

OpenCV Error: The function/feature is not implemented (called functionality is disabled for The current build or platform) in throw_no_cuda, file D: \ opencv3.4 \ OpenCV \ OpenCV \ sources \ modules \ core \ include \ opencv2/core/private cuda. HPP, line 113

CodePudding user response:

Inside the CMake set without using CUDA, recompiling the whole OpenCV

CodePudding user response:

I am going to use cuda, when closed, it doesn't make sense, do you mean don't need to recompile cuda part also can use?

CodePudding user response:

Opencv3.4 version, whether to fully support the cuda9.1, this need to verify, although compiled passed, however, the realization of the opencv version, should be to look ahead from the date of the release of cuda stable releases, is the highest version of opencv can support cuda, cuda set in, if the ground is not impossible, abnormal operation,,
The above comments represent only individual opinions and is for reference only

CodePudding user response:

Inside the CMake set using CUDA, recompiling the whole OpenCV

CodePudding user response:

Back to the third floor, thank you for your advice, don't have time before the whole this, then I put cuda9.1 back to the same cuda8.0 triggered the same exception, 4th floor suggested I didn't understand the I have compiled a lot of time well, many times

CodePudding user response:

Take a look at your DLL lib what can find, I think is a configuration problem