How to use OpenCV3.1 with NVIDIA Nsight?

Robotics & Edge Computing Jetson & Embedded Systems Jetson TX1
6

Hi,

Sorry for keeping you waiting.
I tested Nsight with opencv today and it works well.

Here is my procedure:

1. Libraries and headers:

scp -r ubuntu@[device IP]:[opencv source]/release/lib [host folder]
scp -r ubuntu@[device IP]:/usr/local/include [host folder]
scp ubuntu@[device IP]:/usr/local/cuda-8.0/lib64/libcudart.so.8.0.34 [host folder]/lib
scp ubuntu@[device IP]:/usr/lib/aarch64-linux-gnu/libtbb.so.2 [host folder]/lib
scp ubuntu@[device IP]:/lib/aarch64-linux-gnu/libz.so.1.2.8 [host folder]/lib
sudo ln -s [host folder]/lib/libcudart.so.8.0.34 [host folder]/lib/libcudart.so
sudo ln -s [host folder]/lib/libtbb.so.2 [host folder]/lib/libtbb.so
sudo ln -s [host folder]/lib/libz.so.1.2.8 [host folder]/lib/libz.so

2. Create Nsight project
File → New → CUDA C/C++ Project
Empty Project → Generate PTX code=5.3, Generate GPU code=5.3 → CPU Architecture=AArch64

3. Add configuration (right click → properties)
Build → Settings → Tool Settings
NVCC Compiler → Includes → Include paths → Add “[host folder]/include”
NVCC Linker → Libraries → Libraries → Add “opencv_core”, “cudart”, “tbb”, “z”
NVCC Linker → Libraries → Library search path → Add “[host folder]/lib”

4. Add your source, here is my test code

#include<opencv2/core.hpp>

int main(void)
{
    cv::Mat img;
    return 0;
}

5. Click hammer for compiling.

6. Run → Run as → Remote C/C++ Application