Best Small Hardware to use with OpenCV for Position Tracking

asked 2017-08-22 01:47:16 -0500

Lukaso gravatar image

Hello,

I am running some tests with OpenCV on a Raspberry PI with using Charuco to do some basic camera position tracking. Tho it works, unfortunately on the PI I only get around 3 to 5 fps on that which results in unwanted inaccuary in between the frames.

What is the best small hardware to use for OpenCV at around the size of the PI? Can be a little bit bigger, but not like Mini ITX or so. Is Charuco the most efficient way to track the camera position or are there any other methods within OpenCV that could do that more efficiently on small hardware?

I basically just need the camera position as many times per second as possible, optimal would be above 20 times per second, so that tracking stays accurate.

I hope somebody has already done something similar and can give me some advice.

Thank you! Lukas

edit retag flag offensive close merge delete

Comments

3~5 fps on free run or including processing ? did you try multithreading ?

Ziri gravatar imageZiri ( 2017-08-22 04:10:25 -0500 )edit

including processing, yes. Reading from Camera, processing in opencv, sending out the position via serial. I did not yet try, as I expect it to at best double, which is still far from the targeted 20+ fps.

Lukaso gravatar imageLukaso ( 2017-08-22 04:18:29 -0500 )edit

Did you check out Nvidia Jetson TX2/TX1. The basic HW without peropherals has a size of a credit card.

Ice_T02 gravatar imageIce_T02 ( 2017-08-22 08:29:44 -0500 )edit