Jetson Convenience Script

by FREE WING

http://www.neko.ne.jp/~freewing/

Donation Button

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose 1.6.0 with cuDNN 8.0

Caffe doesn't support cuDNN v8.0 .
So require disable USE_CUDNN .
https://forums.developer.nvidia.com/t/jetpack-4-4-l4t-r32-4-3-production-release/140870/21

OpenPose 1.7.0 support cuDNN 8.0

for Jetson Nano / Jetson Xavier NX Developer Kit

NVIDIA Jetson Nano Developer Kit (4GB model)
https://www.nvidia.com/ja-jp/autonomous-machines/embedded-systems/jetson-nano-developer-kit/
NVIDIA Jetson Xavier NX Developer Kit
https://www.nvidia.com/ja-jp/autonomous-machines/embedded-systems/jetson-xavier-nx/

* not tested Jetson Nano 2GB Developer Kit

NVIDIA JetPack

2022/04
JetPack 5.0 PR I will be Support
JetPack 4.6.1 not Support (Script will not work)

JetPack	L4T Ver.	/etc/nv_tegra_release	Jetson_Convenience_Script Support
4.6 PR	L4T 32.6.1	R32 (release), REVISION: 6.1	add Support JetPack 4.6
4.5.1 PR	L4T 32.5.1	R32 (release), REVISION: 5.1	Archived tag:JetPack_4.5.1
4.5 PR	L4T 32.5	R32 (release), REVISION: 5.0	Archived tag:JetPack_4.5.1
4.4.1 PR	L4T 32.4.4	R32 (release), REVISION: 4.4	Archived tag:JetPack_4.4.1
4.4 PR	L4T 32.4.3	R32 (release), REVISION: 4.3	Archived tag:JetPack_4.4.1
4.4 DP	L4T 32.4.2	R32 (release), REVISION: 4.2	Archived tag:JetPack_4.4.1
4.3 PR	L4T 32.3.1	xxx	Archived tag:JetPack_4.4.1

2021/08 JetPack 4.6 PR Production Release
https://developer.nvidia.com/embedded/jetpack
2021/02 JetPack 4.5.1 PR Production Release
https://developer.nvidia.com/jetpack-sdk-451-archive
JetPack Archive
https://developer.nvidia.com/embedded/jetpack-archive

Jetson Nano / Jetson Xavier NX HEADLESS MODE Setup

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_xavier_nx_developer_kit_headless_mode_setup/

NVIDIA Jetson Nano、Jetson Xavier NX Developer Kit HEADLESS MODE Setup
You can use a Jetson Xavier NX Developer Kit in headless mode, that is , without attaching a display .

* Caution *
Need to Disconnect Display Cable or Power off Display .
If JETSON Detects the Display , It will not go into HEADLESS MODE Setup .

_
_

Jetson WiFi Setup via Terminal Command Line

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_setup_wifi_connection_nmcli/

SSID='WIFI-SSID'
PASSWORD='PLAIN-PASSWORD'

sudo nmcli device wifi connect $SSID password $PASSWORD

sudo nmcli con add type wifi con-name $SSID ifname wlan0 ssid $SSID
sudo nmcli con modify $SSID wifi-sec.key-mgmt wpa-psk
sudo nmcli con modify $SSID wifi-sec.psk $PASSWORD
sudo nmcli con up $SSID
sleep 5
sudo nmcli con up $SSID

sudo nmcli dev wifi rescan
nmcli dev wifi list

sudo ifconfig wlan0 up
ifconfig wlan0
ifconfig -s wlan0

Jetson Nano / Jetson Xavier NX more Memory !!

disable X Window System X11 GUI
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_disable_gui_more_memory/
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_2020_initialize/

Xavier NX	used	free
Before	used 516M	free 6.2G
After	used 291M	free 7.0G

Nano	used	free
Before	used 459M	free 2.7G
After	used 229M	free 3.3G

# Before used 516M, free 6.2G
free -h
#               total        used        free      shared  buff/cache   available
# Mem:           7.6G        516M        6.2G         29M        874M        6.9G
# Swap:          3.8G          0B        3.8G

systemctl get-default
# graphical.target

# disable X Window System X11 GUI
sudo systemctl set-default multi-user.target

# Reboot
sudo reboot

# After used 291M, free 7.0G
free -h
#               total        used        free      shared  buff/cache   available
# Mem:           7.6G        291M        7.0G         19M        321M        7.1G
# Swap:          3.8G          0B        3.8G

# Enable X Window System X11 GUI
sudo systemctl set-default graphical.target
sudo reboot

# Temporarily disable X Window System X11 GUI
sudo systemctl isolate multi-user

# Temporarily disable X Window System X11 GUI
cd
bash ./Jetson_Convenience_Script/JetPack/more_Memory_disable_GUI.sh

Jetson Nano / Jetson Xavier NX Make swap 6GB and Disable nvzramconfig zram swap

Make swap 6GB and Disable nvzramconfig zram swap

cd
bash ./Jetson_Convenience_Script/JetPack/make_swap_6gb.sh

Jetson Xavier NX M.2 NVMe SSD

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_xavier_nx_install_m2_nvme_ssd/

# Format SSD Device to GPT Linux filesystem
cd
bash ./Jetson_Convenience_Script/Xavier_NX_M2_NVMe_SSD/format_m2_nvme_ssd.sh

# Mount SSD Device Add UUID to fstab
bash ./Jetson_Convenience_Script/Xavier_NX_M2_NVMe_SSD/mount_fstab_m2_nvme_ssd.sh

sudo reboot

df -h
# /dev/nvme0n1p1  469G   73M  445G   1% /jetson_ssd

Jetson Xavier NX Booting from M.2 NVMe SSD

Jetson Xavier NX - Run from SSD
https://www.jetsonhacks.com/2020/05/29/jetson-xavier-nx-run-from-ssd/

# Format SSD Device to GPT Linux filesystem
cd
bash ./Jetson_Convenience_Script/Xavier_NX_M2_NVMe_SSD/format_m2_nvme_ssd.sh

# Jetson Xavier NX Booting from M.2 NVMe SSD
bash ./Jetson_Convenience_Script/Xavier_NX_M2_NVMe_SSD/boot_from_m2_nvme_ssd.sh

sudo reboot

# Boot from M.2 NVMe SSD
mount | grep nvme
# Mount NVMe to / root
# /dev/nvme0n1p1 on /

# Disable NVMe Boot
# (Change to Boot from SD-Card)
sudo wipefs /dev/nvme0n1
sudo wipefs --all --force /dev/nvme0n1
sudo reboot

Disk Speed Benchmark M.2 NVMe SSD vs SD-Card

Device	Seq. Read
M.2 NVMe SSD	1576.45 MB/sec
SD-Card	85.49 MB/sec

# Speed Benchmark M.2 NVMe SSD vs SD-Card
sudo apt install -y hdparm

sudo hdparm -t /dev/nvme0n1
# /dev/nvme0n1:
#  Timing buffered disk reads: 4730 MB in  3.00 seconds = 1576.45 MB/sec

sudo hdparm -t /dev/mmcblk0
# /dev/mmcblk0:
#  Timing buffered disk reads: 258 MB in  3.02 seconds =  85.49 MB/sec

Dump Disk Sector

Wipe Bulk Erase Disk Sector

Wipe Erase Disk Partition

more Information to
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_xavier_nx_install_m2_nvme_ssd/

Jetson Nano / Jetson Xavier NX initialize

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_xavier_nx_2020_initialize/
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_2020_initialize/

# Auto detect Nano or Xavier
cd
git clone https://github.com/FREEWING-JP/Jetson_Convenience_Script --depth 1
cd
bash ./Jetson_Convenience_Script/JetPack/1st_jetson_initialize.sh

source .bashrc
 or 
sudo reboot

# sudo visudo
sudo visudo
Defaults        env_reset, timestamp_timeout=-1
 or 
echo 'Defaults env_reset, timestamp_timeout=-1' | sudo EDITOR='tee -a' visudo

Optional deb package

cd
git clone https://github.com/FREEWING-JP/Jetson_Convenience_Script 00_deb -b 00_deb
mv ./00_deb/00_deb/* ./00_deb/
# */
ls -l ./00_deb
---
OpenCV-4.4.0-aarch64-dev.deb
OpenCV-4.5.1-aarch64-dev.deb
bazel_372.zip
cmake-3.17.5-Linux-aarch64.deb
cmake-3.19.4-Linux-aarch64.deb
mediapipe-0.8-cp36-cp36m-linux_aarch64.whl

CMake 3.19.4/ CMake 3.17.5

https://github.com/Kitware/CMake
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_build_newest_cmake/

# CMake 3.19.4
# for Build OpenPose
cd
bash ./Jetson_Convenience_Script/CMake/inst_CMake.sh

# Create .deb install package
bash ./Jetson_Convenience_Script/CMake/create_CMake_deb.sh

# CMake 3.17.5
# for Build OpenPose
cd
bash ./Jetson_Convenience_Script/CMake/inst_CMake_3175.sh

libjpeg-turbo 2.0.5 (libjpeg v8)

https://github.com/libjpeg-turbo/libjpeg-turbo
http://lfsbookja.osdn.jp/BLFS/svn-ja/general/libjpeg.html
-D WITH_JPEG8=ON This switch enables compatibility with libjpeg version 8 .
https://libjpeg-turbo.org/About/TurboJPEG
"libjpeg-turbo" != "TurboJPEG"

cd
bash ./Jetson_Convenience_Script/libjpeg-turbo/inst_libjpeg-turbo_205.sh

OpenBLAS develop

https://github.com/xianyi/OpenBLAS

cd
bash ./Jetson_Convenience_Script/OpenBLAS/inst_OpenBLAS.sh

Bazel 4.0.0/ Bazel 3.7.2/ Bazel 3.5.0

https://bazel.build/
https://github.com/bazelbuild/bazel/tree/3.5.0

# Bazel 4.0.0
cd
bash ./Jetson_Convenience_Script/Bazel/inst_Bazel_400.sh

# Bazel 3.7.2
cd
bash ./Jetson_Convenience_Script/Bazel/inst_Bazel_372.sh

# Bazel 3.5.0
cd
bash ./Jetson_Convenience_Script/Bazel/inst_Bazel_350.sh

OpenCV 3.x

https://github.com/opencv/opencv
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_build_opencv_3410/

# OpenCV 3.4.10
cd
bash ./Jetson_Convenience_Script/OpenCV/inst_OpenCV3410.sh

# OpenCV 3.4.9
cd
bash ./Jetson_Convenience_Script/OpenCV/inst_OpenCV349.sh

OpenCV 4.5.1 with cuDNN 8.0, GStreamer, V4L Video4Linux

cd
bash ./Jetson_Convenience_Script/OpenCV/inst_OpenCV451.sh

# Create .deb install package
bash ./Jetson_Convenience_Script/OpenCV/create_OpenCV_deb.sh

OpenCV 4.4.0 with cuDNN 8.0, GStreamer, V4L Video4Linux

cd
bash ./Jetson_Convenience_Script/OpenCV/inst_OpenCV440.sh

Caffe master

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose with cuDNN 8.0

https://github.com/BVLC/caffe
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_build_caffe_google_deep_dream/

JetPack	USE_CUDNN=1	USE_CUDNN=0
4.5 or later	NG	OK
4.4.1 PR	NG	OK
4.4 PR	NG	OK
4.4 DP	OK	OK
4.3 PR	OK	OK

# with OpenCV 3.x (JetPack 4.2)
# with OpenCV 4.x (JetPack 4.3 or 4.4)
# Auto detect OpenCV 3.x/ 4.x with OpenCV 4.x patch
# support JetPack 4.4 production release disable cuDNN
cd
bash ./Jetson_Convenience_Script/Caffe/inst_Caffe.sh

# Special adapted for OpenCV 4.1 and Python 3.6+
# https://github.com/Qengineering/caffe
# Install OpenCV 4.1.2 and Caffe on Ubuntu 18.04 for Python 3
# https://qengineering.eu/install-caffe-on-ubuntu-18.04-with-opencv-4.1.html
# with OpenCV 4.x
cd
bash ./Jetson_Convenience_Script/Caffe/inst_Caffe_Qengineering.sh

# Caffe installation on Xavier
# https://forums.developer.nvidia.com/t/caffe-installation-on-xavier/67730
# with OpenCV 3.x
cd
bash ./Jetson_Convenience_Script/Caffe/inst_Caffe_NVIDIA.sh

Caffe Deep Dreamer (Google's DeepDream)

https://github.com/kesara/deepdreamer
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_build_caffe_google_deep_dream/

# Auto detect Python 2/ Python 3 with Python 2 patch
cd
bash ./Jetson_Convenience_Script/Caffe/inst_DeepDreamer.sh

_

OpenPose v1.7.0

Support cuDNN 8.0 and OpenCV 4.x

JetPack	OpenPose builtin Caffe	external Caffe	external NVIDIA Caffe v0.17.4
4.5	OK (with cuDNN)	No support	OK (with cuDNN)

# Auto detect JetPack 4.3 or 4.4 or 4.5
# Auto detect OpenCV 4.x for Build OpenPose's Caffe
# Require CMake Version 3.12 or above
# support JetPack 4.4 production release with cuDNN 8.0
cd
bash ./Jetson_Convenience_Script/OpenPose/inst_OpenPose.sh

OpenPose v1.6.0

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose with cuDNN 8.0

https://github.com/CMU-Perceptual-Computing-Lab/openpose
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_2020_build_openpose/
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_xavier_nx_2020_build_openpose/

JetPack	OpenPose builtin Caffe	external Caffe	external NVIDIA Caffe v0.17.3
4.5 or later	OK (without cuDNN)	NG (without cuDNN)	OK (without cuDNN)
4.4.1 PR	OK (without cuDNN)	NG (without cuDNN)	OK (without cuDNN)
4.4 PR	OK (without cuDNN)	NG (without cuDNN)	OK (without cuDNN)
4.4 DP	OK	NG	OK
4.3 PR	OK	NG	OK

# Auto detect JetPack 4.3 or 4.4
# Auto detect OpenCV 3.x/ 4.x for Build OpenPose's Caffe
# external Caffe version should be 0.17.3 (ex. OpenPose internal/ NVIDIA Caffe)
# Require CMake Version 3.12 or above
# support JetPack 4.4 production release without cuDNN 8.0
cd
bash ./Jetson_Convenience_Script/OpenPose/inst_OpenPose_160.sh

OpenPose v1.6.0 Detecting human skeleton NVIDIA Jetson Xavier NX JetPack 4.4 YouTube https://youtu.be/TyokrHR_S_8

OpenPose Benchmark Comparison Jetson Xavier NX vs Jetson Nano

JetPack 4.4 PR + OpenPose v1.6.0 USE_CUDNN=0

net_resolution	Nano	Xavier NX
240x-1	126 sec	108 sec
320x-1	206 sec	116 sec
480x-1	456 sec	137 sec
512x-1	Killed	154 sec
640x-1	Killed	243 sec
none	Killed	254 sec

Movie Spec:
Resolution: 1280x720 px
Frame rate: 25 fps
Duration: 14 sec
Total frame: 350 frame

Original Movie from Pixabay:
https://pixabay.com/videos/id-1643/

Command Line:
./build/examples/openpose/openpose.bin --video 'India - 1643.mp4' --display 0 --model_folder ./models --write_video India_out.mp4 --net_resolution 240x-1
#  --net_resolution xxxxx

tf-pose-estimation master

https://github.com/ildoonet/tf-pose-estimation
https://github.com/gsethi2409/tf-pose-estimation
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_jetpack_tf_pose_estimation_setup/

# with TensorFlow v1.x
cd
bash ./Jetson_Convenience_Script/tf-pose-estimation/inst_tf-pose-estimation.sh

# with TensorFlow v2.x
# https://github.com/gsethi2409/tf-pose-estimation
cd
bash ./Jetson_Convenience_Script/tf-pose-estimation/inst_tf-pose-estimation_tf_v2.sh

_

StyleGAN

https://github.com/NVlabs/stylegan
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_tensorflow_stylegan/
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_tensorflow_stylegan_pretty_anime_face/

# with TensorFlow v1.x
cd
bash ./Jetson_Convenience_Script/StyleGAN/inst_StyleGAN.sh

StyleGAN2

https://github.com/NVlabs/stylegan2
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_tensorflow_stylegan2/

# with TensorFlow v1.x
cd
bash ./Jetson_Convenience_Script/StyleGAN2/inst_StyleGAN2.sh

_

trt_pose

https://github.com/NVIDIA-AI-IOT/trt_pose
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_build_trt_pose/

# Require Pytorch
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_PyTorch_v1_8_Python3.sh

# Require JupyterLab or Jupyter Notebook
bash ./Jetson_Convenience_Script/Jupyter/inst_Jupyter_Notebook.sh

# trt_pose
cd
bash ./Jetson_Convenience_Script/trt_pose/inst_trt_pose.sh

trt_pose_hand

https://github.com/NVIDIA-AI-IOT/trt_pose_hand
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_build_trt_pose_hand/

# Require Pytorch
# Require JupyterLab or Jupyter Notebook
# Require trt_pose
# trt_pose_hand
cd
bash ./Jetson_Convenience_Script/trt_pose_hand/inst_trt_pose_hand.sh

trt_pose_demo

A Demo Application for NVIDIA TensorRT Pose Estimation
https://github.com/MACNICA-CLAVIS-NV/trt_pose_demo

# Require Pytorch
# Require trt_pose

# trt_pose_demo
cd
bash ./Jetson_Convenience_Script/trt_pose_demo/inst_trt_pose_demo.sh

trt_pose_demo Jetson Xavier A Demo Application for NVIDIA TensorRT Pose Estimation YouTube https://youtu.be/Rr4mOH-6f9g

OpenPifPaf 0.12.5 / 0.12.2 / 0.11.9

https://github.com/vita-epfl/openpifpaf
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_install_openpifpaf/

# OpenPifPaf 0.12.5
# Require Pytorch >= 1.7.1
cd
bash ./Jetson_Convenience_Script/openpifpaf/inst_openpifpaf_0125.sh

# OpenPifPaf 0.12.2
# Require Pytorch >= 1.7.1
cd
bash ./Jetson_Convenience_Script/openpifpaf/inst_openpifpaf_0122.sh

# OpenPifPaf 0.11.9
# Require Pytorch >= 1.3.1
cd
bash ./Jetson_Convenience_Script/openpifpaf/inst_openpifpaf_0119.sh

# OpenPifPaf Image Human Pose Estimation
python3 -m openpifpaf.predict ashinari_369878.jpg --long-edge 641 --image-min-dpi=200 --show-file-extension=jpg --image-output

# OpenPifPaf Movie Human Pose Estimation
VIDEO_FILE="'India - 1643.mp4'"
python3 -m openpifpaf.video --source=$VIDEO_FILE --long-edge 641 --video-output video_$VIDEO_FILE

写真素材足成 > 人物 > ナベ散歩
http://www.ashinari.com/2012/09/13-369878.php?

OpenPifPaf with NVIDIA Jetson Nano developer kit Detecting human skeleton from video YouTube https://youtu.be/TTfejnA4yxA

MediaPipe v0.8.6 Python package

https://github.com/google/mediapipe
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_build_mediapipe_python_package/

# MediaPipe v0.8.6 Python package
cd
git clone https://github.com/FREEWING-JP/Jetson_Convenience_Script 00_deb -b 00_deb
mv ./00_deb/00_deb/* ./00_deb/
# */
ls -l ./00_deb

# opencv-contrib-python
# https://pypi.org/project/opencv-contrib-python/3.4.15.55/
opencv_version
# 3.4.15-dev
pip3 install opencv-contrib-python==3.4.15.55

# Install MediaPipe Python package
pip3 install ./00_deb/mediapipe-0.8-cp36-cp36m-linux_aarch64.whl

NVIDIA Jetson Nano Google MediaPipe v0.8.6 Hands Sample YouTube https://youtu.be/QagT6eqdENQ
YouTube https://youtu.be/UZcJR_URUI4
YouTube https://youtu.be/VJucvNfgifg
YouTube https://youtu.be/OyLYgCbBVLI
YouTube https://youtu.be/NaCX8V0H_A0

Tokyo2020-Pictogram-using-MediaPipe

https://github.com/Kazuhito00/Tokyo2020-Pictogram-using-MediaPipe
NVIDIA Jetsonで Tokyo2020オリンピック風のピクトグラムを表示を動かす方法
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_tokyo2020_pictogram_using_mediapipe/

Pictogram-san App

https://github.com/tommy19970714/pictogram-san
NVIDIA Jetsonで東京五輪のピクトグラムさんになれるゲーム Pictogram Challengeを動かす方法
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_tokyo2020_pictogram_san_using_mediapipe/

NVIDIA Caffe v0.17.4

Support cuDNN 8.0 and OpenCV 4.x

JetPack	USE_CUDNN=1	USE_CUDNN=0
4.5 or later	OK	OK

# with OpenCV 4.x (JetPack 4.3 or 4.4 or 4.5)
# Auto detect OpenCV 4.x
# support JetPack 4.4 production release enable cuDNN
cd
bash ./Jetson_Convenience_Script/NV_Caffe/inst_NV_Caffe.sh

NVIDIA Caffe v0.17.3

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose with cuDNN 8.0

https://github.com/nvidia/caffe
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_build_nvcaffe_google_deep_dream/

JetPack	USE_CUDNN=1	USE_CUDNN=0
4.5 or later	NG	OK
4.4.1 PR	NG	OK
4.4 PR	NG	OK
4.4 DP	OK	OK
4.3 PR	OK	OK

# with OpenCV 3.x (JetPack 4.2)
# with OpenCV 4.x (JetPack 4.3 or 4.4)
# Auto detect OpenCV 3.x/ 4.x with OpenCV 4.x patch
# support JetPack 4.4 production release disable cuDNN
cd
bash ./Jetson_Convenience_Script/NV_Caffe/inst_NV_Caffe_0173.sh

# Auto detect Nano or Xavier
cd
bash ./Jetson_Convenience_Script/NV_FFmpeg/inst_NV_FFmpeg.sh

# 2020/09 disable x265
# ffmpeg --enable-libx265
# ERROR: x265 not found using pkg-config

TensorFlow

https://github.com/tensorflow/tensorflow
https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform/index.html
https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform-release-notes/tf-jetson-rel.html
Official TensorFlow for Jetson Nano!
https://forums.developer.nvidia.com/t/official-tensorflow-for-jetson-nano/71770
Official TensorFlow for Jetson AGX XavierNX
https://forums.developer.nvidia.com/t/official-tensorflow-for-jetson-agx-xaviernx/141306
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_jetpack_tensorflow_setup/

TensorFlow Current Version

https://developer.download.nvidia.com/compute/redist/jp/v46/tensorflow/

2021-12-07 for JetPack 4.6
tensorflow-1.15.5+nv21.12-cp36-cp36m-linux_aarch64.whl 220MB
tensorflow-2.6.2+nv21.12-cp36-cp36m-linux_aarch64.whl 302MB

TensorFlow v1.x

# TensorFlow v1.15.5
# Auto detect JetPack 4.5/ 4.6
cd
bash ./Jetson_Convenience_Script/TensorFlow/inst_tf1.sh

TensorFlow v2.x

# TensorFlow v2.5.0
# Auto detect JetPack 4.5/ 4.6
cd
bash ./Jetson_Convenience_Script/TensorFlow/inst_tf2.sh

Build TensorFlow v2.4.1

# Build TensorFlow v2.4.1
# Require Bazel 3.7.2
cd
bash ./Jetson_Convenience_Script/TensorFlow/build_TensorFlow_v2_4_1.sh

Pytorch

https://forums.developer.nvidia.com/t/pytorch-for-jetson-version-1-6-0-now-available/72048

JetPack	Pytorch 1.4.0	1.5.0	1.6.0	1.7.0	1.8.0	1.9.0
4.6 or later	--	--	--	--	--	OK
4.5 PR	--	--	OK	OK	OK	OK
4.4.1 PR	--	--	OK	OK	OK	--
4.4 PR	--	--	OK	OK	OK	--
4.4 DP	OK	OK	OK	--	--	--
4.3 PR	OK	OK	OK	--	--	--

# Pytorch v1.9.0 / torchvision v0.10.0 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_PyTorch_v1_9_Python3.sh

# Pytorch v1.8.0 / torchvision v0.9.0 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_PyTorch_v1_8_Python3.sh

# Pytorch v1.7.0 / torchvision v0.8.1 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_PyTorch_v1_7_Python3.sh

# Pytorch v1.6.0 / torchvision v0.7.0 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_PyTorch_v1_6_Python3.sh

# Pytorch v1.5.0 / torchvision v0.6.0 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_PyTorch_v1_5_Python3.sh

# Pytorch v1.4.0 / torchvision v0.5.0 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_PyTorch_v1_4_Python3.sh

Build PyTorch 1.7.1

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_build_pytorch_1_7_1/

Jetson Model	Build time
Xavier NX	360min
Nano	700min

# Temporarily disable X Window System X11 GUI
cd
bash ./Jetson_Convenience_Script/JetPack/more_Memory_disable_GUI.sh

# gcc 8.4.0
# Recommend use gcc 8.4

# Pytorch v1.7.1 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/build_PyTorch_v1_7_1_Python3.sh

# reboot
sudo reboot

# torchvision v0.8.2 / Python 3.6
cd
bash ./Jetson_Convenience_Script/PyTorch/inst_torchvision_v0_8_2.sh

NVIDIA DeepStream SDK

https://developer.nvidia.com/deepstream-sdk
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_install_deep_stream_sdk/

# NVIDIA DeepStream 5.1 SDK
cd
bash ./Jetson_Convenience_Script/DeepStream/inst_deepstream_51.sh
source .bashrc

# NVIDIA DeepStream 5.0 SDK
cd
bash ./Jetson_Convenience_Script/DeepStream/inst_deepstream_50.sh
source .bashrc

DeepStream Human Pose Estimation

https://github.com/NVIDIA-AI-IOT/deepstream_pose_estimation
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_install_deep_stream_pose_estimation/

Jetson Model	deepstream_pose_estimation_config.txt	Work
Xavier NX	workspace-size=3000	OK
Nano	workspace-size=3000	Killed
Nano	workspace-size=2000	OK

# DeepStream Human Pose Estimation
cd
bash ./Jetson_Convenience_Script/DeepStream/inst_deepstream_pose_estimation.sh

# Jetson Nano patch
# deepstream_pose_estimation_config.txt
# Change workspace-size=3000 to 2000
sed -i 's/^workspace-size=3000/workspace-size=2000/' deepstream_pose_estimation_config.txt

NVIDIA Jetson Xavier NX DeepStream Human Pose Estimation Sample YouTube https://youtu.be/GN1HvM__4gY

JupyterLab or Jupyter Notebook

https://jupyter.org/

# JupyterLab (New) include Jupyter Notebook
# jupyter lab      : 3.0.9
cd
bash ./Jetson_Convenience_Script/Jupyter/inst_JupyterLab.sh

# Jupyter Notebook (classic)
# jupyter-notebook : 6.2.0
cd
bash ./Jetson_Convenience_Script/Jupyter/inst_Jupyter_Notebook.sh

DATA BASE

Redis
Memcached
MongoDB

Redis 6.0.8

https://redis.io/

# Redis
cd
bash ./Jetson_Convenience_Script/Redis/inst_Redis.sh

Memcached 1.6.7

https://memcached.org/
https://github.com/memcached/memcached

# Memcached
cd
bash ./Jetson_Convenience_Script/Memcached/inst_Memcached.sh

MongoDB 3.x / 4.x

https://www.mongodb.com/
https://github.com/mongodb/mongo
32GB SD-Card is Not Enough to Build MongoDB

# MongoDB 3.6.3-0ubuntu1.1 all
apt search mongodb
sudo apt install -y mongodb

# MongoDB 3.4.14
# https://github.com/mongodb/mongo/tree/r3.4.14
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_3414.sh

# MongoDB 3.6.8
# https://github.com/mongodb/mongo/tree/r3.6.8
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_368.sh

# MongoDB 3.6.20
# https://github.com/mongodb/mongo/tree/r3.6.20
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_3620.sh

# MongoDB 4.2.0
# https://github.com/mongodb/mongo/tree/r4.2.0
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_420.sh

# MongoDB 4.2.9
# https://github.com/mongodb/mongo/tree/r4.2.9
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_429.sh

# MongoDB 4.4.1
# https://github.com/mongodb/mongo/tree/r4.4.1
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_441.sh

# MongoDB 4.7.0
# https://github.com/mongodb/mongo/tree/r4.7.0
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_470.sh

MongoDB Tools 3.x / 4.x

https://docs.mongodb.com/tools/
https://github.com/mongodb/mongo-tools

# MongoDB Tools master
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_Tools.sh

# MongoDB Tools r3.6.20
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_Tools.sh r3.6.20

# MongoDB Tools r4.0.20
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_Tools.sh r4.0.20

# MongoDB Tools r4.2.10
cd
bash ./Jetson_Convenience_Script/MongoDB/inst_MongoDB_Tools.sh r4.2.10

Visual Studio Code

https://github.com/Microsoft/vscode
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_build_visual_studio_code_oss/

# Visual Studio Code 1.53.2
# for Jetson Xavier
cd
bash ./Jetson_Convenience_Script/Visual_Studio_Code/inst_Visual_Studio_Code_1532.sh

# for Jetson Nano
# Download .deb package
# https://code.visualstudio.com/#alt-downloads

# Visual Studio Code 1.52.1
# for Jetson Xavier
cd
bash ./Jetson_Convenience_Script/Visual_Studio_Code/inst_Visual_Studio_Code_1521.sh

# Visual Studio Code 1.47.2
# for Jetson Xavier
cd
bash ./Jetson_Convenience_Script/Visual_Studio_Code/inst_Visual_Studio_Code_1472.sh

# Visual Studio Code 1.35.0
# for Jetson Nano
cd
bash ./Jetson_Convenience_Script/Visual_Studio_Code/inst_Visual_Studio_Code_1350.sh

Vino VNC Server

https://developer.nvidia.com/embedded/learn/tutorials/vnc-setup
https://gitlab.gnome.org/GNOME/vino/
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_enable_vino_vnc_server_headless_mode/

cd
bash ./Jetson_Convenience_Script/Vino_VNC/inst_Vino_VNC.sh

# Password = password
# gsettings set org.gnome.Vino vnc-password $(echo -n 'password'|base64)

_

Benchmark

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_benchmark_full_load/

# UnixBench byte-unixbench
# https://github.com/kdlucas/byte-unixbench
cd
bash ./Jetson_Convenience_Script/Benchmark/inst_UnixBench.sh

# Benchmarks Targeted for Jetson Xavier NX (Using GPU+2DLA)
# https://github.com/NVIDIA-AI-IOT/jetson_benchmarks
cd
bash ./Jetson_Convenience_Script/Benchmark/inst_jetson_benchmarks.sh

Jetson stats

https://github.com/rbonghi/jetson_stats

# Install
sudo -H pip install -U jetson-stats
sudo reboot

Jetson Hello AI World

https://developer.nvidia.com/embedded/twodaystoademo
https://github.com/dusty-nv/jetson-inference
http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_nano_sample_application/

# Building the Project from Source
# https://github.com/dusty-nv/jetson-inference/blob/master/docs/building-repo.md
cd
bash ./Jetson_Convenience_Script/Jetson_Hello_AI_World/inst_Jetson_Hello_AI_World.sh

Ubuntu Desktop GNOME Screensaver

http://www.neko.ne.jp/~freewing/raspberry_pi/ubuntsu_desktop_gonome_disable_screen_saver/

# Disable Ubuntu Desktop GNOME Screensaver
gsettings set org.gnome.desktop.lockdown disable-lock-screen true
gsettings set org.gnome.desktop.screensaver lock-enabled false
gsettings set org.gnome.desktop.screensaver ubuntu-lock-on-suspend false
gsettings set org.gnome.desktop.screensaver idle-activation-enabled false

# Enable Ubuntu Desktop GNOME Screensaver (Unit: sec)
gsettings set org.gnome.desktop.lockdown disable-lock-screen false
gsettings set org.gnome.desktop.session idle-delay 0
gsettings set org.gnome.desktop.session idle-delay $((15*60)) && \
gsettings set org.gnome.desktop.screensaver lock-delay 5 && \
gsettings set org.gnome.desktop.screensaver lock-enabled true

Build Python 3.9.2/ Python 3.8.8/ Python 3.7.10

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_build_python_3_7_10/

# Build Python 3.9.2
cd
bash ./Jetson_Convenience_Script/Python/build_Python_392.sh

# Build Python 3.8.8
cd
bash ./Jetson_Convenience_Script/Python/build_Python_388.sh

# Build Python 3.7.10
cd
bash ./Jetson_Convenience_Script/Python/build_Python_3710.sh

gcc 8.4.0

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_install_gcc_8_4_0/

cd
bash ./Jetson_Convenience_Script/gcc/inst_gcc_840.sh

Clang 10

http://www.neko.ne.jp/~freewing/raspberry_pi/nvidia_jetson_install_gcc_8_4_0/

cd
bash ./Jetson_Convenience_Script/Clang/inst_Clang_10.sh

FWinSdCardImager Windows SD-Card Image Read Write App.

http://www.neko.ne.jp/~freewing/software/windows_sd_card_imager/
https://www.vector.co.jp/soft/winnt/util/se520996.html

Name		Name	Last commit message	Last commit date
Latest commit History 307 Commits
Bazel		Bazel
Benchmark		Benchmark
CMake		CMake
Caffe		Caffe
Clang		Clang
DeepStream		DeepStream
JetPack		JetPack
Jetson_Hello_AI_World		Jetson_Hello_AI_World
Jupyter		Jupyter
Memcached		Memcached
MongoDB		MongoDB
NV_Caffe		NV_Caffe
NV_FFmpeg		NV_FFmpeg
OpenBLAS		OpenBLAS
OpenCV		OpenCV
OpenPose		OpenPose
PyTorch		PyTorch
Python		Python
Redis		Redis
StyleGAN		StyleGAN
StyleGAN2		StyleGAN2
TensorFlow		TensorFlow
Vino_VNC		Vino_VNC
Visual_Studio_Code		Visual_Studio_Code
Xavier_NX_M2_NVMe_SSD		Xavier_NX_M2_NVMe_SSD
gcc		gcc
libjpeg-turbo		libjpeg-turbo
llvm		llvm
openpifpaf		openpifpaf
tf-pose-estimation		tf-pose-estimation
trt_pose		trt_pose
trt_pose_demo		trt_pose_demo
trt_pose_hand		trt_pose_hand
LICENSE		LICENSE
README.md		README.md
bell.sh		bell.sh

License

FREEWING-JP/Jetson_Convenience_Script

Folders and files

Latest commit

History

Repository files navigation

Jetson Convenience Script

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose 1.6.0 with cuDNN 8.0

for Jetson Nano / Jetson Xavier NX Developer Kit

* not tested Jetson Nano 2GB Developer Kit

NVIDIA JetPack

Jetson Nano / Jetson Xavier NX HEADLESS MODE Setup

Jetson WiFi Setup via Terminal Command Line

Jetson Nano / Jetson Xavier NX more Memory !!

Jetson Nano / Jetson Xavier NX Make swap 6GB and Disable nvzramconfig zram swap

Jetson Xavier NX M.2 NVMe SSD

Jetson Xavier NX Booting from M.2 NVMe SSD

Disk Speed Benchmark M.2 NVMe SSD vs SD-Card

Dump Disk Sector

Wipe Bulk Erase Disk Sector

Wipe Erase Disk Partition

Jetson Nano / Jetson Xavier NX initialize

Optional deb package

CMake 3.19.4/ CMake 3.17.5

libjpeg-turbo 2.0.5 (libjpeg v8)

OpenBLAS develop

Bazel 4.0.0/ Bazel 3.7.2/ Bazel 3.5.0

OpenCV 3.x

OpenCV 4.5.1 with cuDNN 8.0, GStreamer, V4L Video4Linux

OpenCV 4.4.0 with cuDNN 8.0, GStreamer, V4L Video4Linux

Caffe master

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose with cuDNN 8.0

Caffe Deep Dreamer (Google's DeepDream)

OpenPose v1.7.0

OpenPose v1.6.0

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose with cuDNN 8.0

OpenPose Benchmark Comparison Jetson Xavier NX vs Jetson Nano

tf-pose-estimation master

StyleGAN

StyleGAN2

trt_pose

trt_pose_hand

trt_pose_demo

OpenPifPaf 0.12.5 / 0.12.2 / 0.11.9

MediaPipe v0.8.6 Python package

Tokyo2020-Pictogram-using-MediaPipe

Pictogram-san App

NVIDIA Caffe v0.17.4

NVIDIA Caffe v0.17.3

JetPack 4.4 production release L4T 32.4.3 Can't build Caffe and OpenPose with cuDNN 8.0

NVIDIA FFmpeg for Jetson Nano

NVIDIA FFmpeg for Jetson Xavier NX master

NVIDIA FFmpeg for Jetson Nano / Jetson Xavier NX

TensorFlow

TensorFlow Current Version

TensorFlow v1.x

TensorFlow v2.x

Build TensorFlow v2.4.1

Pytorch

Build PyTorch 1.7.1

NVIDIA DeepStream SDK

DeepStream Human Pose Estimation

JupyterLab or Jupyter Notebook

DATA BASE

Redis 6.0.8

Memcached 1.6.7

MongoDB 3.x / 4.x

MongoDB Tools 3.x / 4.x

Visual Studio Code

Vino VNC Server

Benchmark

Jetson stats

Jetson Hello AI World

Ubuntu Desktop GNOME Screensaver

Build Python 3.9.2/ Python 3.8.8/ Python 3.7.10

gcc 8.4.0

Clang 10

FWinSdCardImager Windows SD-Card Image Read Write App.

About

Topics

Packages