NO WAY TO USE GPU - TORCH NOT DETECTING GPU - CUDA 12.6 - JP 6.1 ORIN NANO

Robotics & Edge Computing Jetson Systems Jetson Orin Nano
1

I have tried to install PyTorch torch-vision and torch-audio for my Jetson Orin Nano
I have used the pre-built one itself from the below website,

but im stil getting the same errors when im checking

image
image1006×257 52.6 KB

and this too
meghana@meghana-desktop:~$ python -c “import torch; print(torch.version)”
2.10.0+cpu

why is it still cpu?

4

Hi,

You can find the cuDSS library below:

The CPU PyTorch package is listed in the apt so you can check if any package (ex., TorchVision) has installed it to meet the dependencies.

Thanks.

6

image
image1208×620 133 KB

image
image1188×414 80.4 KB

now GPU is installed, why is it still not working?
i downloaded the pre-trained model yolov8s from the Ultralytics and I’m trying to train this model with my dataset to detect if the person is wearing a safety suit or not, basically a ppe detection.
so i downloaded the dataset from the Kaggle and below is the data.yaml file for the dataset

image
image888×294 41.8 KB

how do i go about it?

7

You have multiple Python environments. You installed PyTorch with CUDA in yoloenv but installed Ultralytics in the local system environment (see the path of Python packages in your logs; they are different for PyTorch with CUDA vs..Ultralytics). You should uninstall Ultralytics from system environment and install it to yoloenv.

deactivate
pip uninstall ultralytics -y
source yoloenv/bin/activate
pip install ultralytics
8

There is no update from you for a period, assuming this is not an issue anymore.
Hence, we are closing this topic. If need further support, please open a new one.
Thanks ~0324

Hi,

Could you try to remove the torch-2.10.0+cpu on your environment so Ultralytics script won’t load it by default?

Thanks.