error when running docker container with dustynv/jetson-inference:r32.6.1 (built from nvidia:l4t-pytorch)

I just use the image dustynv/jetson-inference:r32.6.1 , which is built from nvidia:l4t-pytorch, to be my base image to build docker image. Hoever when the image built, I get into the container ans execute my python program, but got the following error:

import jetson.inference
  File "/usr/lib/python3.6/dist-packages/jetson/inference/", line 5, in <module>
    import jetson.utils
  File "/usr/lib/python3.6/dist-packages/jetson/utils/", line 4, in <module>
    from jetson_utils_python import *
ImportError: cannot open shared object file: No such file or directory

Any idea on this?

hi @user6348

I guess this error persist when Cuda and other packages are not installed,
Can you please once confirm what version of CUDA, CUDNN, and TensorRT are you using or whether is it installed on your jetson device. basically when you flash SD card using Jetpack it comes preloaded.
once can you please refer this as well NVIDIA Container Runtime on Jetson · NVIDIA/nvidia-docker Wiki · GitHub hope this solves your issue

Thank you

1 Like


Which JetPack do you use?
Could you check if the system is setup with the JetPack 4.6 (r32.6.1)?


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