1 About the update source
The information provided by the seller does not recommend changing the source, the reason: Jetson TX2 NX uses the Ubuntu 18.04.2 LTS system with aarch64 architecture, which is different from the Ubuntu system with AMD architecture, and no perfect domestic source has been found, so it is not recommended to change the source.
I didn't change the source, but the speed of installing some libraries is ok, first do the following
sudo apt-get update
as the picture shows:
Then, execute the following command
sudo apt-get full-upgrade
as the picture shows:
During the command execution process, this window pops up, I directly choose No to continue:
You can enter commands to view CUDA
nvcc -V
as the picture shows:
To check Opencv, first enter the command
python
Then enter the following python code
import cv2
cv2.__version__
You can see the Opencv version number, and then enter the exit command
exit()
as the picture shows:
Test the onboard camera, write commands
nvgstcapture-1.0
as the picture shows:
Enter q directly and press Enter to turn off the camera:
2 Jetson TX2 NX Install TensorFlow GPU
Install the pip command as follows
sudo apt-get install python3-pip python3-dev
The installation may fail due to network reasons, try several times, as shown in the figure:
To upgrade pip, enter the command
python3 -m pip install --upgrade pip
as the picture shows:
Install the numpy package, enter the command
sudo apt-get install python3-numpy
as the picture shows:
Install the scipy package, enter the command
sudo apt-get install python3-scipy
as the picture shows:
Install the pandas package, enter the command
sudo apt-get install python3-pandas
as the picture shows:
Install the matplotlib package, enter the command
sudo apt-get install python3-matplotlib
as the picture shows:
Install the sklearn package, enter the command
sudo apt-get install python3-sklearn
as the picture shows:
Install the libhdf5-serial-dev package, enter the command
sudo apt-get install libhdf5-serial-dev
as the picture shows:
Install the hdf5-tools package, enter the command
sudo apt-get install hdf5-tools
as the picture shows:
Install the libhdf5-dev package, enter the command
sudo apt-get install libhdf5-dev
as the picture shows:
Install the zlib1g-dev package, enter the command
sudo apt-get install zlib1g-dev
as the picture shows:
Install the zip package, enter the command
sudo apt-get install zip
as the picture shows:
Install the libjpeg8-dev package, enter the command
sudo apt-get install libjpeg8-dev
as the picture shows:
Install the liblapack-dev package, enter the command
sudo apt-get install liblapack-dev
as the picture shows:
Install the libblas-dev package, enter the command
sudo apt-get install libblas-dev
as the picture shows:
Install the gfortran package, enter the command
sudo apt-get install gfortran
as the picture shows:
Install TensorFlow GPU offline, copy the installation file to Jetson TX2 NX directly using a USB flash drive, and then execute the installation command
pip3 install tensorflow-2.1.0+nv20.4-cp36-cp36m-linux_aarch64.whl
The following errors may occur. The reason for the error is that the connection is interrupted when downloading some packages:
Solution: Since some software packages will be installed during the installation process, in order to improve the download speed and stability, now change the source for pip
Pip permanently changes the source, enter the following command to create .pip
mkdir ~/.pip
Then create a pip.conf file and enter the command
vim ~/.pip/pip.conf
After entering, press ito enter edit mode and copy the Tsinghua source below
[global]
index-url = https://pypi.tuna.tsinghua.edu.cn/simple
trusted-host = pypi.tuna.tsinghua.edu.cn
Then press ESCexit, press :wqsave to exit
Then execute the installation command, as shown in the figure:
An error is reported when installing the h5py package, as shown in the figure:
Tried many solutions on the Internet, none of which worked, and finally installed the h5py package by entering the following command
sudo apt install python3-h5py
as the picture shows:
Keras is an advanced neural network API written in Python, the installation command is as follows
sudo pip3 install keras
as the picture shows:
I hope this article is helpful to everyone. If there is anything wrong with the above, please correct me.
Sharing determines the height, and learning widens the gap