Yolov8 live camera inference framerate is too slow

Community Help
1

I am currently on Raspberry Pi 5 with the RPI OS 64-bit and I am running an inference with a live camera with the default weight “yolov8n.pt”.
I tried exporting the weight to other formats (ncnn, onnx, edgetpu) but I am never going over 3fps.

The code I am using is :

from ultralytics import YOLO
model = YOLO(“yolov8n.pt”)
results = model(source=0, show=True, conf=0.3, save=True, workers=4)

I tried with yolov8x too but same result.

Would you have some advices to make the frame rate faster ?

Thanks in advance !