I am using OpenCV 3.0 with VS2012 C++/CLI on a Win 8.1 64 bit machine.
In addition, I have an NVidia GTX 760 with CUDA 7.0.
I am trying to detect SURF keypoints but only within an ROI.
I got this working in non-GPU code and am now trying to do the same thing using the GPU implementation.
In order to do that, I understand that I first need to create a mask filled with a background value (0) and then specify an ROI within the mask and fill that with another value (255). I then pass this mask in as part of the SURF call.
The following code is supposed to create the mask and fill it with zeros.
cuda::GpuMat tMask(dImages->d_gpuGrayFrame->size(), dImages->d_gpuGrayFrame->type());
tMask.setTo(Scalar::all(0));
This code causes an immediate crash. I have also tried specifying the type as CV_8U and that still causes a crash.
Any idea on what is wrong? I am hoping that @Vladislav-Vinogradov might be monitoring as he appears to be a GPU guru.
Thanks,
James
↧