Jetson Xavier NX OpenCV & TensorFlow

#Pythonの環境構築

##Jetpackのバージョンを確認
$sudo apt-cache show nvidia-jetpack
Package: nvidia-jetpack
Version: 4.4-b144
Architecture: arm64
Maintainer: NVIDIA Corporation

##Python3のバージョンを確認
$python3 --version
Python 3.6.9

##pipのインストール
$sudo apt install python3-pip

##virtualenvのインストール

$ mkdir tf
$ cd tf
$ sudo python3 -m pip install virtualenv
$ virtualenv env

env/bin/activateに以下の２行を追加する。

OLD_PYTHONPATH=$PYTHONPATH
export PYTHONPATH="/usr/local/lib/python3.6/dist-packages:/usr/lib/python3/dist-packages:/usr/lib/python3.6/dist-packages"

env/bin/postdeactivateに以下の２行を追加する。

export PYTHONPATH=$OLD_PYTHONPATH
unset OLD_PYTHONPATH

##OpenCVのインストール
JetsonHacks

$ git clone https://github.com/jetsonhacks/buildOpenCVXavier.git
$ cd buildOpenCVXavier
$ git checkout v1.0
$ ./buildOpenCV.sh

###build Examples

これはエラーになる
$ g++ -o gstreamer_view -Wall -std=c++11 gstreamer_view.cpp $(pkg-config –libs opencv)

なのでこっち
$ g++ -o gstreamer_view -Wall -std=c++11 gstreamer_view.cpp `pkg-config --cflags --libs opencv`
$ sudo ldconfig 
$ ./gstreamer_view

##tensorflowのインストール

(env) nx@nx-desktop:~/tf $ sudo apt-get install -y libhdf5-dev libc-ares-dev libeigen3-dev
(env) nx@nx-desktop:~/tf $ sudo apt-get install -y libhdf5-serial-dev hdf5-tools zlib1g-dev zip libjpeg8-dev liblapack-dev libblas-dev gfortran
(env) nx@nx-desktop:~/tf $ sudo apt-get install -y openmpi-bin libopenmpi-dev
(env) nx@nx-desktop:~/tf $ sudo apt-get install -y libatlas-base-dev
(env) nx@nx-desktop:~/tf $ python3 -m pip uninstall tensorflow
(env) nx@nx-desktop:~/tf $ pip3 install -U numpy==1.16.1 future==0.17.1 grpcio absl-py py-cpuinfo psutil portpicker six mock==3.0.5 requests gast==0.2.2 h5py==2.9.0 astor termcolor protobuf keras_preprocessing==1.0.5 keras_applications==1.0.8 wrapt google-pasta setuptools testresources scipy==1.4.1

###JP_VERSION
The major and minor version of JetPack you are using, such as 42 for JetPack 4.2.2 or 33 for JetPack 3.3.1.

###TF_VERSION
The released version of TensorFlow, for example, 1.13.1.

###NV_VERSION
The monthly NVIDIA container version of TensorFlow, for example, 19.01.

$ pip3 install --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v44 tensorflow==$TF_VERSION+nv$NV_VERSION


e.g
2系
$ pip3 install --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v44 tensorflow==2.1.0+nv20.04
1系
$ pip3 install --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v44 tensorflow==1.51.2+nv20.04

###インストール後の確認
仮想環境を一旦終了して再度起動する。

(env) pi@raspberrypi:~/tf_pi $ deactivate 
pi@raspberrypi:~/tf_pi $ source env/bin/activate 
(env) pi@raspberrypi:~/tf_pi $ python3 
>>>import tensorflow
>>>

エラーにならずにインポートできればOK

##参考
https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform-release-notes/tf-jetson-rel.html#tf-jetson-rel
https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform/index.html
https://forums.developer.nvidia.com/t/official-tensorflow-for-jetson-agx-xavier/65523
https://www.tensorflow.org/tutorials/quickstart/beginner