Issue running ONNXRUNTIME in Jetson

I followed the steps in follow tutorial inside a Jetson Xavier:

but, when I try to import onnxruntime in python3 I have follow error:

  • import onnxruntime

/usr/local/lib/python3.6/dist-packages/onnxruntime/capi/_pybind_state.py:14: UserWarning: Cannot load onnxruntime.capi. Error: ‘libcublas.so.10: cannot open shared object file: No such file or directory’.
warnings.warn(“Cannot load onnxruntime.capi. Error: ‘{0}’.”.format(str(e)))
Traceback (most recent call last):
File “”, line 1, in
File “/usr/local/lib/python3.6/dist-packages/onnxruntime/ init .py”, line 13, in
from onnxruntime.capi._pybind_state import get_all_providers, get_available_providers, get_device, set_seed,
ImportError: cannot import name ‘get_all_providers’

I’m completely stuck in this error and I don’t understand the root cause, I tried also using other different versions of onnxruntime, but always have same issue
What should be the issue?

Hi,

Have you set up the CUDA library on Jetson?
This can be done by installing all the components from JetPack.

Thanks.

I’m using a docker thats should have already installed CUDA libraries by it’s own, so I suppose that don’t need to install drivers on OS. The driver should be installed in OS Kernel too?

The issue was a lot of missing dependencies in the base jetpack docker. After we added them, now it works correctly

Hi,

Thanks for the update.

Please note that the base image (l4t-base) only contains L4T OS.
You can find other containers with the CUDA or TensorRT installed:

CUDA: NVIDIA L4T CUDA | NVIDIA NGC
TensorRT: NVIDIA L4T TensorRT | NVIDIA NGC

Thanks.