Jetson nano performance in object tracking

Hi there,
I have a question about jetson nano board using. I watched lots of video about motion tracking on jetson nano. And i saw nearly all of them shows low performance although above 20 FPS was claimed. Most of them was seen so laggy and showed nearly 3-5 FPS(opencv-python). I was thinking that nano cards could do more stable image processing with gpu hosting. I need this for my personal project. Is there any way to increase this performance? Could this be caused by cpu usage instead of gpu usage? I need some guidance on this i want to get this card but i am confused.

Hi,

Sorry for the late update.

The performance depends on the algorithm/library you used.
It’s more recommended to use our library since it has otpimized for Jetson Nano.

You can find the performance of a baseline pipeline on Jetson Nano below:
https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_Performance.html#jetson-nano

Thanks.