CUDA won't run with with our yolov8 object detection

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
1

Need urgent help as i was not able to run my yolov8 object detection on my jetson orin nano super development kit.

jetpack l4t 36.4.4
cuda = 12.6

please help

2

*** Please note that this reply is generated by LLM automatically ***

I’d be happy to help you with running YOLOv8 object detection on your Jetson Orin Nano Super development kit.

Firstly, I’d like to confirm that you are using JetPack 6.2/Jetson Linux 36.4.4, CUDA 12.6, and the Jetson Orin Nano Super development kit. This is a great setup for running AI models like YOLOv8.

To troubleshoot the issue, could you please provide more details about the error you are encountering? For example, are you getting any error messages, and if so, what do they say? Additionally, have you tried running any other AI models on your Jetson Orin Nano Super, and if so, did they work successfully?

Assuming you have installed the necessary dependencies, including CUDA, cuDNN, and OpenCV, here are some potential steps to resolve the issue:

  1. Verify CUDA and cuDNN versions: Ensure that you have the correct versions of CUDA and cuDNN installed on your Jetson Orin Nano Super. You can check the versions by running the following commands:
nvcc --version
nvidia-smi
  1. Check OpenCV version: Verify that you have the correct version of OpenCV installed. You can check the version by running the following command:
pkg-config --modversion opencv4
  1. Update JetPack and CUDA: If you are using an older version of JetPack or CUDA, try updating to the latest versions. You can do this by running the following commands:
sudo apt update
sudo apt install nvidia-jetpack
sudo apt install cuda-toolkit-12-6
  1. Rebuild and reinstall OpenCV: If you are using a pre-built version of OpenCV, try rebuilding and reinstalling it from source. You can do this by running the following commands:
git clone https://github.com/opencv/opencv.git
cd opencv
mkdir build
cd build
cmake ..
make
sudo make install
  1. Verify YOLOv8 installation: Ensure that you have correctly installed YOLOv8 on your Jetson Orin Nano Super. You can do this by running the following command:
git clone https://github.com/ultralytics/yolov8.git
cd yolov8
pip install -r requirements.txt
  1. Run YOLOv8 with CUDA: Try running YOLOv8 with CUDA enabled. You can do this by running the following command:
python detect.py --weights yolov8s.pt --source 0 --device 0

If you are still encountering issues, please provide more details about the error you are seeing, and I’ll do my best to help you troubleshoot the problem.

In comparison to other Jetson modules, the Jetson Orin Nano Super has a more powerful GPU and more memory, which should make it well-suited for running AI models like YOLOv8. However, the troubleshooting steps I provided above should be applicable to other Jetson modules as well.

I hope this helps! Let me know if you have any further questions or if there’s anything else I can help you with.

*** Please give the thumbs up if you get answers you like. Or provide feedback to help us improve the answer. ***

5

Hi,

Please find the tutorial for YOLOv8 below:

Thanks.