How to solve segmentation fault error on raspberry pi 4

Question

I am currently trying to use yolov8 to perform object detection on the raspberry pi 4. I have installed ultralytics and other necessary packages but whenever i run the code on the terminal it says "segmentation fault". Here is the code that I am trying to run:

from ultralytics import YOLO
model = YOLO("yolov8n.pt")
model.predict(source=0, show=True)

How can I fix this error?

• Thorbjørn Ravn Andersen: Segmentation fault means that some native code has tried to manipulate memory unavailable to it. Typically this is because of unhandled invalid input data. Are you certain your code provides all the parameters needed and that they are as expected. If you follow a tutorial it may be outdated.
• MatsK: Solve this is something yolov8 can do, NOT Raspberry Pi.
• Eme Obioh: It is showinh me this error message: "WARNING ⚠️ inference results will accumulate in RAM unless stream=True is passed, causing potential out-of-memory errors for large sources or long-running streams and videos." I have included the stream=True but i still keep getting the message
• goldilocks: Might be useful if you posted the full text of the seg fault.

Answer 1

Downgrading torch and torchvision worked for me.

pip uninstall torch torchvision
pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2

You can read more about this issue at https://github.com/ultralytics/ultralytics/issues/5274

Answer 2

Try this!

from ultralytics import YOLO
model = YOLO("yolov8n.pt")
results = model.predict(source=0, stream=True)
for r in results:
 boxes = r.boxes # Bounding box coordinates
 masks = r.masks # Segmentation masks
 probs = r.probs # Class probabilities
r.render()

• Eme Obioh: @goldilocks Thanks for the reply. I tried what you suggested and this is what i get: 1/1: 0... Success ✅ (inf frames of shape 640x480 at 30.00 FPS) Segmentation fault The segmentation fault error is still there and the video does not even show
• goldilocks: I didn't suggest anything, I just formatted the code in Sam's answer.