There's a wheel for jetpack 4.6 of TF2.5 here https://elinux.org/Jetson_Zoo#ONNX_Runtime

Thank you for the image!
After flashing and installing xrdp, I connect to my jetson nano B01 and the screen freezes as shown below.
Do you have some hints that can help with this problem please?

I notice that using the transforms provided by PyTorch for their pretrained models in a Jupyter Notebook (in VSCode), the results were wildly incorrect. I managed to pinpoint the issue on the normalization step. In fact, even just calculating the channel-wise mean of a tensor gives wrong results. Surprisingly, the results are correct (that is, they match their Numpy counterpart) if the tensor is moved to the GPU first or if not Jupyter is involved.

Has anyone notice similar behavior ?

My pip list:

When running pip install -r requirements.txt it is stuck at
"Building wheels for collected packages: grpcio"
Is this normal?

Building wheels for collected packages: grpcio
Building wheel for grpcio (setup.py) ... /
I am using the unclocked version, after an apt-get upgrade
Thanks

can I install it in external SSD and run it from there ?

I use SDXC 128gb but after i flash system found it as 32GB ..
Can I solve this problem?

Thank you

Hi,

I wanted to know if anyone has tried to use jetson inference on this image? According to the official Github of Jetson Inference, this image includes all the requirements except that I am not sure if Tensorrt works on this image.

Thanks in advance !

Hey!
I'm very glad to see your image that already have everything I need.
But I can't download it.

Hello,

I installed your custom image (which by the way works very well for image recognition and ros2 foxy thank you very much) and I tried to use PWM 32 and 33. I have enabled it with Jetson-IO, rebooted but I can't get a pwm signal on either pin (32 and 33). I have no errors in my python code (a really simple one, it was working on jetpack 4.6 (official one) with the same activation process). Does anyone have any idea how I can solve this problem?

Thanks in advance,

Hi
this is output of opencv versioin and gstream is disable :

can you build in python3 with gstreamer on ?

Cannot import tensorrt using python3

I see /usr/local/cuda-10.2, but "nvcc --version" has no response. Thanx in advance. But is very useful, that I can finally use Torch, my target. Thanx a lot. I wish to use .cu to compile in vscode,

Please update cmake to a newer version >= 3.25.0.

The image seems not to work with the Jetson Nano 2GB. It is stuck on the logo screen.

I have problem when try implementation of https://nlp.seas.harvard.edu/annotated-transformer/
on torch 1.12.0 torchdata need torch 1.11.0,
but on torch 1.11.0 torchtext got error

''' >>> import torchtext
Traceback (most recent call last):
File "", line 1, in
File "/home/jetson/.local/lib/python3.8/site-packages/torchtext/init.py", line 9, in
from . import vocab
File "/home/jetson/.local/lib/python3.8/site-packages/torchtext/vocab/init.py", line 11, in
from .vocab_factory import (
File "/home/jetson/.local/lib/python3.8/site-packages/torchtext/vocab/vocab_factory.py", line 4, in
from torchtext._torchtext import (
ImportError: /home/jetson/.local/lib/python3.8/site-packages/torchtext/_torchtext.so: undefined symbol: _ZNK5torch3jit6MethodclESt6vectorIN3c106IValueESaIS4_EERKSt13unordered_mapISsS4_St4hashISsESt8equal_toISsESaISt4pairIKSsS4_EEE '''

After starting my jetson nano with the new image i tried to connect TeamViewer with my acc. but i only get error:

"
Welcome to TeamViewer configuration.

This short guide helps you to setup TeamViewer on this device. After you have
successfully finished the setup, this device will automatically be available in
your Computers & Contacts.

Connected
Please enter your email / username: test

Please enter your password:
Initiating sign in
There was a connectivity issue. Please check your internet connection and try
again.

"

I reinstalled and checked my Network but nothing worked.

" ping master4.teamviewer.com
PING master4.teamviewer.com(master4.teamviewer.com (2a0b:b580:0:1::4)) 56 data b ytes
64 bytes from master4.teamviewer.com (2a0b:b580:0:1::4): icmp_seq=1 ttl=120 time =9.23 ms
64 bytes from master4.teamviewer.com (2a0b:b580:0:1::4): icmp_seq=2 ttl=120 time =10.5 ms
--- master4.teamviewer.com ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 9.229/9.870/10.512/0.641 ms
"

Hi, can you describe workflow how to create such an image? I would like to create similar images for Xavier NX or other Jetson devices, and I used your image on nano successfully :)

Hi,

When looking at jtop I can see opencv is compiled with CUDA yes.

When running a script It is defaulting to CPU then running a sample script shows opencv is not compiled with CUDA.

Is it a enviroment path issue?

Or is the OPENCV 4.8 version not actually built with CUDA?

Any help in understanding would be great

I can't load Ubuntu 20.04 image, with OpenCV 4.5.3, TensorFlow 2.4.1 and PyTorch 1.9.0.

Clean setup with latest image, brings an crash with the opencv library under python

python3

import cv2
Illegal instruction (core dumped)

Any ideas ?

I just recently flashed my SanDisk Extreme microSDXC 128GB memory card with the image. It works, but it only uses up to 32GB of space.

Do you have any plans to upgrade this image to Ubuntu 22.04?

@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
@ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY!
Someone could be eavesdropping on you right now (man-in-the-middle attack)!
It is also possible that a host key has just been changed.
The fingerprint for the ECDSA key sent by the remote host is
SHA256:3r+XqpYUBuE8osiMj2x71Za7JEYwbolhCBpd//M8vFY.
Please contact your system administrator.
Add correct host key in /home/pravin/.ssh/known_hosts to get rid of this message.
Offending ECDSA key in /home/pravin/.ssh/known_hosts:14
remove with:
ssh-keygen -f "/home/pravin/.ssh/known_hosts" -R "192.168.226.77"
ECDSA host key for 192.168.226.77 has changed and you have requested strict checking.
Host key verification failed.

what should i do to use SSH key as usual?

How can I get an image for the jetson nano ubuntu 20 without opencv preinstalled. I need to use a specific version of opencv but I can not figure out how to uninstall it and I am not able to have multiple versions of opencv installed. I think that it was installed using these instructions. However the ubuntu image doesn't contain the original directory that opencv was built in so I can't uninstall it. I was able to find the shared libraries in the /usr/lib/aarch64-linux-gnu directory however I am not sure how to safely uninstall them.

I am running a the most recent image without any modifications.

Hi,

I'm trying to install an edge microserver for thingworx. And there are some weird things going on in your image.
The main problem is, that the /lib/ld-linux.so.3 interpreter is pointing to a non existing armhf file, even the jetson nano architecture is arm64. What did install it?

I'm trying to execute the following application:

If I'm running it, it says No such file or directory

Hope you can answer my question and help. Thanks alot!

one day, I switch to text mode with the following command:
sudo systemctl set-default multi-user.target
since this day, connecting through ssh give the following message:
'DISPLAY' environment variable not set... skipping surface info
and these strange lines in dmesg:
[ 42.248674] tegra-xusb 70090000.xusb: WARN: buffer overrun event on endpoint
[ 42.255820] tegra-xusb 70090000.xusb: WARN: buffer overrun event on endpoint
[ 42.263670] tegra-xusb 70090000.xusb: WARN: buffer overrun event on endpoint

I've tried to fill the DISPLAY environment variable with :0.0 => Same result

I never see this issue with the standard jetson linux version

as I am using a WEBCAM i met strange behaviours sometime with a lot of these strange messages.

Any idea?

Dear all, first of all congratulations for the great work, it has been of great help in my project due to the ambiguous version of python that NVIDIA maintains in jetson nano, I would like to know how to make a CSI camera (raspberry or IMX) work with OpenCV in this system operational, follow this workflow https://github.com/JetsonHacksNano/CSI-Camera and got failure, stay tuned, regards.

Hello
There are problems when loading the kernel.
Perhaps you can help solve it.
Tried installing the last two images, the result is the same. Used 32 GB and 64 GB SD card.
The only thing is probably that Jetson Nano is for Chinese market.

Is it possible to installed custom os with MMC SSD instead of sd card to production board Nano device?

Error found when loading /etc/profile:

WARNING: lavapipe is not a conformant vulkan implementation, testing use only.

As a result the session will not be configured correctly.
You should fix the problem as soon as feasible.

warning 1:
/usr/bin/xauth: file /home/jetson/.Xauthority does not exist

I want to use python3.8
So I upgrade my jetson nano ubuntu from 18.04 to 20.04
the code which I can run successfully in 18.04 have error in 20.04
How can I solve these problems? Thanks a lot
I had check there is no problem in hardware device

code:
from jetcam.csi_camera import CSICamera
camera = CSICamera(width=640, height=480)
image = camera.read()

error:
RuntimeError Traceback (most recent call last)
File /usr/local/lib/python3.8/dist-packages/jetcam-0.0.0-py3.8.egg/jetcam/csi_camera.py:24, in CSICamera.init(self, *args, **kwargs)
23 if not re:
---> 24 raise RuntimeError('Could not read image from camera.')
25 except:

RuntimeError: Could not read image from camera.

During handling of the above exception, another exception occurred:

RuntimeError Traceback (most recent call last)
Cell In[2], line 2
1 from jetcam.csi_camera import CSICamera
----> 2 camera = CSICamera(width=640, height=480)
3 image = camera.read()

File /usr/local/lib/python3.8/dist-packages/jetcam-0.0.0-py3.8.egg/jetcam/csi_camera.py:26, in CSICamera.init(self, *args, **kwargs)
24 raise RuntimeError('Could not read image from camera.')
25 except:
---> 26 raise RuntimeError(
27 'Could not initialize camera. Please see error trace.')
29 atexit.register(self.cap.release)

RuntimeError: Could not initialize camera. Please see error trace.

My board with jetson nano does not have any slot of sd card, so I need an ubuntu 20.04 image of USD flash disk. Can I get support?

Hello!

I was wondering if there is a particular reason why opencv is at version 4.6 instead of 4.8? I was able to successfully install version 4.8 with gstreamer and such, following qengineering tutorials, but was curious if maybe there's a performance difference.

I'm sorry that my question is little bit off but couldn't find a solution.

How to Install qt4 on Jetson Nano (ubuntu 20.04) for Hector Slam

Firstly, I tried

sudo add-apt-repository ppa:rock-core/qt4

this code and tried to install the needed libraries, but in my opinion it couldn't find the libraries again because of the jetson nano's architecture.

Then I tried to download the debian packages from the internet and realized most of them are amd64 packages, I find one package for arm64, but this one needs dependencys as shown below

tried to install downloaded package for 20.04 and arm64

When this gives an error, I tried to download the qt5 then change the version manually, as shown below it doesn't work and says run make install (don't know how to do)

qt version 4.8

I need to install qt4 to run Hector Slam. What should I do any suggestions?

Hello, I was trying to install https://github.com/WCHSoftGroup/ch343ser_linux and hit some issues when running make. I believe this is due to the bare image, as the error I get is:
make[1]: *** /lib/modules/4.9.253/build: No such file or directory.

From what I could tell, it seems the linux-headers-4.9.253 are missing, which is the output of uname -r, but when I try installing headers, all I can get are 5.x. Wondering if anyone has solutions for this!

I tried

sudo apt install python-is-python3
sudo apt install python3-venv
rm -rf venv/
python -m venv venv
source venv/bin/activate
python -m pip install --upgrade pip

and tried
pip install tensorflow==2.10.1
but
python -c 'import tensorflow as tf; print(tf.__version__)'
show me

Traceback (most recent call last):
File "", line 1, in
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/init.py", line 37, in
from tensorflow.python.tools import module_util as _module_util
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/init.py", line 45, in
from tensorflow.python.feature_column import feature_column_lib as feature_column
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/feature_column/feature_column_lib.py", line 18, in
from tensorflow.python.feature_column.feature_column import *
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/feature_column/feature_column.py", line 143, in
from tensorflow.python.layers import base
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/layers/base.py", line 16, in
from tensorflow.python.keras.legacy_tf_layers import base
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/keras/init.py", line 25, in
from tensorflow.python.keras import models
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/keras/models.py", line 22, in
from tensorflow.python.keras.engine import functional
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 32, in
from tensorflow.python.keras.engine import training as training_lib
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/keras/engine/training.py", line 54, in
from tensorflow.python.keras.saving import hdf5_format
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/keras/saving/hdf5_format.py", line 37, in
import h5py
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/h5py/init.py", line 46, in
from ._conv import register_converters as _register_converters
File "h5py/_conv.pyx", line 1, in init h5py._conv
File "h5py/h5t.pyx", line 293, in init h5py.h5t
File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/numpy/init.py", line 320, in getattr
raise AttributeError("module {!r} has no attribute "
AttributeError: module 'numpy' has no attribute 'typeDict'

so I did
pip install numpy==1.20
but python -c 'import tensorflow as tf; print(tf.__version__)'

RuntimeError: module compiled against API version 0x10 but this version of numpy is 0xe
ImportError: numpy.core.multiarray failed to import

The above exception was the direct cause of the following exception:

SystemError: <built-in method __contains__ of dict object at 0x7f7034b640> returned a result with an error set

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/__init__.py", line 37, in <module>
    from tensorflow.python.tools import module_util as _module_util
  File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/__init__.py", line 37, in <module>
    from tensorflow.python.eager import context
  File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/eager/context.py", line 35, in <module>
    from tensorflow.python.client import pywrap_tf_session
  File "/home/hannes/CarND-Behavioral-Cloning-P3/venv/lib/python3.8/site-packages/tensorflow/python/client/pywrap_tf_session.py", line 19, in <module>
    from tensorflow.python.client._pywrap_tf_session import *
ImportError: initialization failed

Does someone knows, how to fix this ?

I'm trying to connect my Jetson Nano using your OS (Ubuntu 20.04) and I tried this tutorial:
https://www.hackster.io/Matchstic/connecting-pixhawk-to-raspberry-pi-and-nvidia-jetson-b263a7

and I got error when tried to execute:
sudo mavproxy.py --master=/dev/ttyTHS1
come up with this error:
Illegal instruction

can you help me with this one? Thank you

We are flashing Jetson Nano 4gb with this image and using Wifi Card: Intel Model 8265NGW. We periodically have an error where the Jetson Nano does not recognize the wifi adaptor. No network settings show up at all and it doesn't not connect to any of its saved wifi networks (they don't show up at all.) To test the issue we updated everything all of the software and then rebooted 10 times. 2/10 times the Jetson had this issue still. We're not sure what's going on here. Thanks for your help!

Steps:

1. Created a new user
2. Added the same groups of user "jetson" to my new user
3. $ python3 -> import torch returns Illegal instruction (core dumped)

torchvision returns the same error
Adding the new user to the "jetson" group doesn't help
There are no problems importing the modules from the "jetson" user

Latest image shows an error after startup with the vulkan driver on Jetson Nano 4GB.

Any ideas, below the message I get after startup:

/build/vulkan-tools-_ynL0i/vulkan-tools-1.2.131.1+dfsg1/vulkaninfo/vulkaninfo.h:477: failed with ERROR_INITIALIZATION_FAILED

Can you help me how to update the Jetpack to the latest version (with L4T 32.7.1)??]

I'm a noob guy hehe.

Thanks in advance!

As the title states, I have a Jetson Nano module with 16GB of eMMC memory instead of an SD card slot. Is it somehow possible to flash this image into memory either with or without NVIDIA's SDKManager?

Your image is great, but it seems to be limited to 32 GB (NVIDIA base installation already done). Do you have a hint, how I could resize the root partition after first start?

dear author,i can not import python package:"tensorrt",is there not include tensorrt pybind in this ubuntu20.04 image?

This github page directs users to the Gdrive page to get the actual Ubuntu 20.0 OS for Nvidia Jetson Nano Dev Kit. But once at the site and trying to download the OS the following error appears:
"Sorry, you can't view or download this file at this time.
Too many users have viewed or downloaded this file recently. Please try accessing the file again later. If the file you are trying to access is particularly large or is shared with many people, it may take up to 24 hours to be able to view or download the file. If you still can't access a file after 24 hours, contact your domain administrator."

This error has consistently been appearing over the past 3 hours.

Does anyone have a work around for this or can the file be placed here, on github?

Hello, I have tried to update my ubuntu on my nano from 18 to 20 but during the upgrading there was a prompt asking me to accept firefox update. I did and now my ubuntu is 22.04 and the nano doesn't work.

What should I do?

÷ have a bluetooth HC-05 module I try to connect it but it doesn't work