YoloV8 Segmentation fault (core dumped)

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
1

Hi,
I have a Jetson Orin Nano that i just set up with the SDK-Manager with everything installed in the newest version. I wanted to try the most basic YoloV8 case and ran
yolo predict model=yolov8n.pt source=‘https://ultralytics.com/images/bus.jpg’
which is the first case shown on the Ultralytics Github and also works on my normal computer. Unfortunatly this simple case returns “Segmentation fault (core dump)” after a few seconds.
Any ideas?

nano@ubuntu:~$ yolo checks
Ultralytics YOLOv8.0.200 🚀 Python-3.8.10 torch-2.1.0 CPU (ARMv8 Processor rev 1 (v8l))
Setup complete ✅ (6 CPUs, 7.3 GB RAM, 67.3/914.4 GB disk)

OS Linux-5.10.120-tegra-aarch64-with-glibc2.29
Environment Linux
Python 3.8.10
Install pip
RAM 7.30 GB
CPU ARMv8 Processor rev 1 (v8l)
CUDA None

matplotlib ✅ 3.7.3>=3.3.0
numpy ✅ 1.24.4>=1.22.2
opencv-python ✅ 4.8.1.78>=4.6.0
pillow ✅ 10.1.0>=7.1.2
pyyaml ✅ 5.3.1>=5.3.1
requests ✅ 2.31.0>=2.23.0
scipy ✅ 1.10.1>=1.4.1
torch ✅ 2.1.0>=1.8.0
torchvision ✅ 0.16.0>=0.9.0
tqdm ✅ 4.66.1>=4.64.0
pandas ✅ 2.0.3>=1.1.4
seaborn ✅ 0.13.0>=0.11.0
psutil ✅ 5.9.6
py-cpuinfo ✅ 9.0.0
nano@ubuntu:~$ yolo predict model=yolov8n.pt source=‘https://ultralytics.com/images/bus.jpg’
Ultralytics YOLOv8.0.200 🚀 Python-3.8.10 torch-2.1.0 CPU (ARMv8 Processor rev 1 (v8l))
YOLOv8n summary (fused): 168 layers, 3151904 parameters, 0 gradients, 8.7 GFLOPs

Found https://ultralytics.com/images/bus.jpg locally at bus.jpg
Segmentation fault (core dump)

3

Hi,

Have you checked the same app on another environment?
Could you try if the sample works on a desktop device first?

Thanks.

4

Hi,

I just tested on my Windows 11 PC, WSL2 (Ubuntu) and an Ubuntu machine. All of those are working:

WSL2:
$ yolo predict model=yolov8n.pt source=‘https://ultralytics.com/images/bus.jpg’
Ultralytics YOLOv8.0.162 🚀 Python-3.10.12 torch-2.0.1+cu117 CPU (AMD Ryzen 7 6800U with Radeon Graphics)
YOLOv8n summary (fused): 168 layers, 3151904 parameters, 0 gradients

Found https://ultralytics.com/images/bus.jpg locally at bus.jpg
image 1/1 /home/ubuntu/bus.jpg: 640x480 4 persons, 1 bus, 1 stop sign, 49.9ms
Speed: 3.0ms preprocess, 49.9ms inference, 1.0ms postprocess per image at shape (1, 3, 640, 480)
Results saved to runs/detect/predict2

W11:
yolo predict model=yolov8n.pt source=‘https://ultralytics.com/images/bus.jpg’
Ultralytics YOLOv8.0.158 Python-3.11.2 torch-2.0.1+cu118 CPU (AMD Ryzen 7 6800U with Radeon Graphics)
YOLOv8n summary (fused): 168 layers, 3151904 parameters, 0 gradients

Found https://ultralytics.com/images/bus.jpg locally at bus.jpg
image 1/1 C:\Users\krause\bus.jpg: 640x480 4 persons, 1 bus, 1 stop sign, 98.6ms
Speed: 2.1ms preprocess, 98.6ms inference, 0.0ms postprocess per image at shape (1, 3, 640, 480)
Results saved to runs\detect\predict5

5

Hi,

Could you check how much memory the app is used?
Since Nano is relatively limited in memory, the error might be cased by a memory shortage.

Thansk

6

YoloMemory
YoloMemory1464×535 229 KB

As far as I can tell there is nothing special to see here. (The small bump at 10s should come from yolo).

7

Hi,

Moving your topic to Orin Nano.

We do get several topics about the segmentation fault issue when running YOLOv8 recently.
Will check it further and update more info with you later.

Thanks.

9

Hi,

Please see the below comment for more info:

Thanks.

10

Thanks, has fixed the problem. (I also had to build torchvision from source according to your guide PyTorch for Jetson and for some reason had to uninstall torchvision 0.16.0 as this was installed for some reason)

12

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.