Deepstack optimization for Jetson Xavier NX

Hello everyone

Face recognition on my Jetson Xavier NX does only 4 FPS despite a fairly small image (1000x600px) and the maximum 20W 6CORE mode. I’m running the latest deepstack:jetpack-2021.06.1 image on jetpack 4.6

Presumably one can optimize deepstack model for Xavier NX. How does one do it?

Cheers

1 Like

Thanks for creating this post. We are continuously working on improving DeepStack’s performance by addressing bottlenecks and optimizing inference where possible. In future releases, you will get better FPS rate than currently obtained from Jetson Xavier.

cc: @john