Missing VPI in Python (VPI 2 DP)

AI & Data Science Computer Vision & Image Processing
1

Hello,
I installed JetPack 5.0 DP via SDK-Manager in Ubuntu 18.04 (as host). SDKM said that all things are installed well, inclusive VPI2. Python environment is prepared with Python3.8 via Anaconda3. But

  1. sudo apt remove python3.8-vpi2
  2. Start to install described in VPI - Vision Programming Interface: Installation
  3. At point 3: “sudo add-apt-repository ‘deb https://repo.download.nvidia.com/jetson/x86_64 bionic r34.1’”. apt send me a warning:

W: Das Laden der konfigurierten Datei »r34.1/binary-i386/Packages« wird übersprungen, da das Depot »https://repo.download.nvidia.com/jetson/x86_64 bionic In Release« die Komponente »r34.1« nicht enthält. (Schreibfehler bei der Angabe der Komponente in sources.list?)

Means the repository dosn’t contain r.34.1.

So installation of python3.8-vpi2 not possible.

Can somewhere help?

3

Hi,

We also provide the packages in the following link:
https://repo.download.nvidia.com/jetson/#Jetpack%205.0

Would you mind installing the VPI python package through the download link to see if it works?

Thanks.

4

Hi,

Thanks for your reply!

By using your link

I took the VPI2 package under x86_64/bionic in version 2.0.14 Result was

Unpacking vpi2-dev (2.0.14) over (2.0.9) ...
vpi2-dev (2.0.14) is set up ...

Unfortunately in python3 the result is:

(jp5.0-py3.8) ~$ python3
Python 3.8.8 (default, Apr 13 2021, 19:58:26) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import vpi
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'vpi'

Are there some folders to copy into anaconda3 folder structure after this installation process? I assume the problem could be that I work inside an anaconda environment and the VPI installer doesn’t recognize this.

Best regards, Frank

5

Hi,

Do you use a virtualenv?
If yes, could you try it with the --system-site-packages to see if it works?

Ex.

$ virtualenv env --system-site-packages

Thanks.

6

Hi,
No, I use Anaconda3 environments.
In this case there is a directory “.conda” inside the user home. And inside this folder is only the file “environments.txt” what contains the directory of Anaconda environments. In my case the Anaconda environments are in “/opt/anaconda3”. This is not inside the user home in order to use this environment by all users. All users have access rights into “/opt/anaconda3” by using a special anaconda user group.

So, my environment for using JetPack5 with Python 3.8 is inside this Anaconda folder: /opt/anaconda3/envs/jp5.0-py3.8

The folder structure inside this environment is:

bin
compiler_compat
conda-meta
doc
etc
include
info
lib
libexec
licensing
man
mkspecs
phrasebooks
plugins
qml
resources
sbin
share
ssl
translations
var
x86_64-conda_cos6-linux-gnu

Does the VPI2 installer use such an environment configuration or is this the reason for this problem?
Best regards, Frank

7

Hi,

Could you try if you can import vpi outside of the Anaconda3 environment?
Thanks.

8

Yes I tried. I set up a normal python3-environment:

virtualenv --python=/usr/bin/python3.8 /opt/python3_environments/py3.8-jp5.0
source /opt/python3_environments/py3.8-jp5.0/bin/activate

Of course, the user has full access into /opt/python3_environments.

Now Jetpack5.0.1 DP reinstalled.

And checked:

(py3.8-jp5.0) (base) herbrand@fwfe15:~$ python3
Python 3.8.13 (default, Apr 19 2022, 00:53:22) 
[GCC 7.5.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import vpi
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'vpi'
>>>
9

Hi,

When setting virtualenv, could you add --system-site-packages to see if it helps?

Thanks.

10

Hi and thanks!

The setup of the virtual environment with --system-site-packages is necessary.

virtualenv --system-site-packages --python=/usr/bin/python3.8 /opt/python3_environments/py3.8-jp5.0

works for me now. But in this case the installation of some Python packages are needed. For >>> import vpi we need also:

pip numpy.

So the VPI import is successfully.

Thanks, Frank

11

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