Hi
I am running yolo26 nativly on my PC, Raspberry Pi and then will expirment with it using Hailo-8. Are there any documents or topics that show what the processing performance rate would be to expect for each platform?
What is the fastest processing we should expect?
Also, can we extract information on how well a frame has been processed for comparison purposes?
You can check the benchmark in the docs:
With YOLO26n exported to OpenVINO on Raspberry Pi 5, you would get about 10FPS.
I am also streaming the video from my Raspberry Pi to my PC for develpment purpose
I am using the YOLO26s.
I am able to achieve a max of boxes=5 | preprocess 2.3 ms, inference 188.3 ms, postprocess 0.4 ms
Does this seem correct?
Also, the camera is able to capture 60Fps, is necessary that you need to analyse each frame. Is it a good practice to analyse alternative frames to save processing resources? As there would be repeated information in frames.
It seems normal. You will have to drop the frames to match inference FPS. Because your inference FPS is about 5.