Not able to install torchvision v0.6.0 on Jetson jetson xavier nx

Hello,

I tried installing pytorch and torchvision on Nvidia Jetson Xavier NX, using the steps given on website PyTorch for Jetson - version 1.8.0 now available. I was able to successfully install pytorch version 1.5 but unable to install torchvision v0.6.0 .

Below are the steps which I have performed to install the torchvision

$ sudo apt-get install libjpeg-dev zlib1g-dev libpython3-dev libavcodec-dev libavformat-dev libswscale-dev

$ git clone --branch v0.6.0 GitHub - pytorch/vision: Datasets, Transforms and Models specific to Computer Vision torchvision # see below for version of torchvision to download

$ cd torchvision

$ export BUILD_VERSION=0.6.0 # where 0.x.0 is the torchvision version

$ python3 setup.py install --user

PFB error received.

Traceback (most recent call last):
File “setup.py”, line 13, in
import torch
File “/home/jetson/archiconda3/envs/MM/lib/python3.6/site-packages/torch/init.py”, line 135, in
_load_global_deps()
File “/home/jetson/archiconda3/envs/MM/lib/python3.6/site-packages/torch/init.py”, line 93, in _load_global_deps
ctypes.CDLL(lib_path, mode=ctypes.RTLD_GLOBAL)
File “/home/jetson/archiconda3/envs/MM/lib/python3.6/ctypes/init.py”, line 348, in init
self._handle = _dlopen(self._name, mode)
OSError: libmpi_cxx.so.20: cannot open shared object file: No such file or directory

CUDA Version**: 10.2
Python Version (if applicable): 3.6
PyTorch Version (if applicable)
: 1.5.1

Could you please help me resolve the error.

Hi,
This looks like a Jetson issue. We recommend you to raise it to the respective platform from the below link

Thanks!

Hi @kshitij.kamlakar, have you installed these dependencies first?

sudo apt-get install libopenblas-base libopenmpi-dev 
sudo apt-get install libjpeg-dev zlib1g-dev libpython3-dev libavcodec-dev libavformat-dev libswscale-dev

Also, I noticed you are installing into archiconda environment, which I haven’t tried, so if the above doesn’t fix the issue you may want to try installing outside of archiconda environment to see if that is related to the problem.