My aim is to use computeAsync function in order to describe vector of cv::KeyPoint in the host memory because I’m only detecting ORB keypoints and then removing outliers before computing descriptors, but I don’t know how I can convert vectorcv::Keypoint type to cuda::GpuMat. Is there any way to compute descriptors in this way?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Fast feature detection in CUDA | 1 | 815 | June 10, 2021 | |
| Asynchronous feature detection not running asynchronously | 3 | 677 | November 10, 2022 | |
| Number of keypoints does not match number of descriptors | 1 | 106 | June 7, 2024 | |
| Opencv cuda stream optimisation | 1 | 1051 | August 18, 2022 | |
| Manually download cv::cuda::GpuMat to cv::Mat using cudaMemcpyAsync | 1 | 94 | August 14, 2024 |