Error during installation of jetcard

[Tzowbiie]

Hello,
i am trying to install jetcard onto my jetson nano. I cloned this repo

https://github.com/NVIDIA-AI-IOT/jetcard/tree/jetpack_4.2.2

Now i receive following error:

Collecting gast
Downloading gast-0.4.0-py3-none-any.whl (9.8 kB)
Collecting h5py
Downloading h5py-3.0.0.tar.gz (370 kB)
|████████████████████████████████| 370 kB 17.6 MB/s
Installing build dependencies ... done
Getting requirements to build wheel ... done
Installing backend dependencies ... error
ERROR: Command errored out with exit status 1:
command: /usr/bin/python3 /usr/local/lib/python3.6/dist-packages/pip install --ignore-installed --no-user --prefix /tmp/pip-build-env-6kveagwx/normal --no-warn-script-location --no-binary :none: --only-binary :none: -i https://pypi.org/simple -- 'Cython>=0.29; python_version < "3.8"' 'Cython>=0.29.14; python_version >= "3.8"' pkgconfig 'numpy==1.17.5; python_version == "3.8"' 'numpy==1.19.3; python_version >= "3.9"' 'numpy==1.14.5; python_version == "3.7"' 'numpy==1.12; python_version == "3.6"'
cwd: None
Complete output (2569 lines):
WARNING: The directory '/home/tobi-jetson/.cache/pip' or its parent directory is not owned or is not writable by the current user. The cache has been disabled. Check the permissions and owner of that directory. If executing pip with sudo, you may want sudo's -H flag.
Ignoring Cython: markers 'python_version >= "3.8"' don't match your environment
Ignoring numpy: markers 'python_version == "3.8"' don't match your environment
Ignoring numpy: markers 'python_version >= "3.9"' don't match your environment

The error message is way longer..

Would be nice if somenone could help. Thanks!

[Tzowbiie]

The error occurs while installing h5py package. No idea how to fix it

[dive76]

Same here. This version has not been adopted to Jetpack 4.4.1.

[mtmal]

Hi all,

I had exactly the same problem with Jetpack 4.4.1. I spent few hours trying various solutions, but finally I went to the JetBot repository, which wiki explains step-by-step how to create SD card image from scratch.
The key step is then installing TensorFlow which requires h5py. If you follow instructions to here, you will find that that setuptools, numpy, and h5py need to be installed with a specified version. Also, the install script from this repository takes the precomputed tensor flow from v43 folder, which is for JetPack 4.3. You want to update that accordingly to your JetPack version just as explained in cited instructions (for 4.4.1 it is v44).

Also:

• tensorflow 2.3 wants gast==0.3.3. It also requires numpy<1.19.0,>=1.16.0. I ended up installing version 1.18.5 and commenting out code which at the beginning upgrades it to the latest version.
• there is no if-check around getting torch-1.6.0-cp36-cp36m-linux_aarch64.whl, so if you may end up executing the script more than once, add it. It will save re-downloading 250MB
• you definitely want to put if-check to test if folder exists where git repositories are being cloned. Otherwise the scripts ends.
• line ~86 of the install script, because it sets ownership of local share folder to default jetson user, which may not exist in your system
• you need to update nodejs to version at least 12. I followed tutorial from here
• swap file will be created in /var/swapfile. I had already swap in a different place and commented that section out

I hope that helps!