Code for YOLOv4 Tiny not using full memory


We have a code for object tracking using YOLOv4 Tiny that we are running on Jetson Nano 2gb as well as Jetson Nano 4gb. We upgraded to 4gb as the FPS on 2gb was 1.7-1.8.
After running the codes on 4gb Jetson Nano we are still getting the same FPS as that of 2gb Jetson Nano.

When I looked at the memory usage of both the Jetsons we got -
1) 2gb Jetson using 1.8gb of the memory available to give 1.8FPS
2) 4gb Jetson using 2.2gb of memory available to give same or lower FPS.

We tried to cap the memory at 3gb, 3.5gb and 4gb but still the code was not using the available memory, it was consistently using 2.2gb and no more

How can we make the code use all the available memory for giving better results in terms of FPS.



The memory usage and performance depends on the implementation.

Do you use the darknet related implementation?
If yes, could you switch to our Deepstream example and give it a try?


This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.