Jetson Nano Object Detection Code running with low performance

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
1

I am trying to run an object detection code which uses opencv, and cvlib libraries created in a python3.11 virtual environment. I get very low frames when I run the code. But when i monitor the jetson stats using jtop I see that only CPU is getting used and GPU is getting used rarely. I have enabled CUDA and its for opencv 4.1.1 but the opencv version that I run in my virtual environment is 4.8.0. Is this the reason the performance is bad? I am new to this field. Help me out! What other things should I keep in mind to get the best performance? Thanks!

2

Hi @ab9hi9 ,
I request you to raise the concern on Jetson platform to get better assistance.

Thank you

4

Hi,

After you built OpenCV with CUDA support, have you switched to using CUDA API?
Below are some examples that use GPU API for your reference:

Thanks.

5

Hey Thanks for that! I am new to this stuff. Are there any videos or tutorials that help explaining the same.

Thanks!!

6

Can I get the link for the same?

7

Hi,

You can find below for some guidance:

Thanks.

9

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.