Is it recommended to use C++ for video analytics using Deepstream compare to Python?

• Hardware Platform: dGPU
• DeepStream Version: 5.0
• TensorRT Version: 7.0.0-1+cuda10.2
• NVIDIA GPU Driver Version: 465.19.01
Hello there,
I have been experimenting with the Deepstream pipeline and ran few checks to see if there’s any performance differences between c++ and python, I noticed that they both consumed almost same amount of memory and also give same amount of FPS, I ran Deepsteram-test1 and deepstream-nvdsanalytics-test app provided by deepstream-python-apps and c++, I couldnt find any substantial difference… any recommendations on when to use c++ over python? We’ve almost built an app in python and now looking for ways to improve the performance.

I would really appreciate any feedback on this.

Thanks in advance.

Most of the process is in plugin level. So there is little performance benifit between different API.