auto-vectorizing a c++ library

edit

Im trying to improve the speed of a ros node I've been using to capture a camera's video, called usb_cam. It currently has a lot of latency between the real-word and its displayed image. I want to use neon to optimize the code to try to eliminate the delay.

Alas, I do not know how I could implement this. I've seen in-code examples but i do not know enough about c++ on how to implement those, and ive seen compiler setting examples but oi do not know how that would be preformed as i launch the node with roslaunch. I managed to find a pretty similar thread here but I don't know if this would suit my case as it's for OpenCV

For reference, my system is:

• ubuntu 16.04 xenial
• ros kinetic kame
• raspberry pi 3b+ (Linux 5.4.y)