Couple of issues - Cuda/easyOCR & SD card backup

Hi all,
I am trying to use easyOCR and I keep getting the following error:
“CUDA not available - defaulting to CPU. Note: This module is much faster with a GPU.”

I am running OpenCV version 4.1.1 and it says compiled CUDA: NO
I am struggling to find what version OpenCV I should be running to make it compatible.

The second long-standing issue is how to clone the image to an SD card. I tried following JetsonHacks’s tutorial but it throws an error after 2 hours of waiting.

Thanks for visiting.

So I found out that torch.cuda.is_available() was returning false. Google search said I need to add environment path. As I opened a new terminal, I was getting error :
~bash: /home/jn/.bash_profile:Permission denied
I know I didn’t login as super user. So I decided to reboot. After rebooting I can’t get in anymore. The login screen looks different. I have no idea what has happened! Please find pictures attached.


How do you install the PyTorch package.
You can find a package that built with GPU support in the below link:


Thanks for getting back AastaLLL. I was in a bit of a pickle trying to get easyOCR up and running. As I mentioned I am unable to get in GUI for some reason and I was also unable to transfer my code across to any device. Primarily because I couldn’t clone SD card. I managed to ssh using putty and initially I was getting the following error:

-bash: cd: username/: Permission denied

I couldn’t access any files from Document folder (cd /home//Document…etc). I found out this error

chown missing operand after#
chown: changing owernshio of '/home/jn/…etc. : Operation not permitted

I think I rectified this by using the following command and luckily managed to transfer data across using WinSCP

chown -R root:users dir

Apparently, PyTorch wasn’t installed and I am still unable to install it for some reason. I am going to flash microSD card and start again. I will make sure PyTorch is installed correctly this time. It doesn’t help much because I cannot enter my credentials to log in to GUI to debug it. Unfortunately, I m not an expert in terminal lol. Maybe I will try creating a virtual environment this time and see if it makes any difference. I am putting some screenshots for future reference. I really appreciate your help. Many thanks.



It seems that the installation requires root authority.
Could you try it with sudo to see if it works?

More, you will need a package built with the same JetPack version.

For example, JetPack 4.6.

$ wget -O torch-1.10.0-cp36-cp36m-linux_aarch64.whl
$ sudo apt-get install python3-pip libopenblas-base libopenmpi-dev 
$ pip3 install Cython
$ pip3 install numpy torch-1.10.0-cp36-cp36m-linux_aarch64.whl

You can find all the details in the below topic:


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