Sorry didn't get any time to try out during the week and get back. I tried it today, installed a fresh 64Bit Bookworm and followed the instructions, ran into some delays, etc but noticed that mem was really an issue. So increased the swap file to 4M and yep works great (of course we still do see latency but it's much better - I will try to squeeze out more perf).The standard 64-bit OS with all the relevant software updates should be all you need, nothing extra.
Also, as my use-case is to have some application/probe that can have the camera installed and when it finds the object it beeps/shows label of object on the LCD type screen attached to the Pi. Hence I may forgo the rendering of the images and bounding boxes - as that's taking CPU to render and display the video.
Any ideas on those lines... also I have an LCD display that can communicate over SPI interface to the Pi - can I use it simultaneously alongside with this camera or will it create SPI bus configs?
MY LCD is a 3rd party from waveshare and uses SPI with this configuration:
https://www.waveshare.com/wiki/1.8inch_ ... connection
My goal is to see if the object detected has a threshold of say 70% or more it should display the label of that object on LCD and mark green/buzzer etc and if not then keep on searching..
Is this something that is possible? I can start with default yolov8 object detection model and classes... (but later have this used to detect custom objects).
And thank you again for your help in providing help and clear solutions.
Statistics: Posted by mhpetiwala — Sat Nov 02, 2024 5:10 pm