Base Install

Use the basimage that comes preinstalled with the most important tools: https://github.com/Qengineering/Jetson-Nano-image

another great source for getting things working on the jestson nano is this repository: https://github.com/step305/pyFaceRec_jetson_nano/blob/master/setup.txt

Test CSI Camera

Install https://github.com/thehapyone/NanoCamera

sudo python3 setup.py install

Run python3 examples/CSI_camera.py

Troubleshooting:

Check if device is recognized:

ls /dev/video0

Check if openCV was built with all necessary option like CUDA and GStreamer support:

python3 import cv2 ``print(cv2.getBuildInformation())´`

If you get "illegal instructions" error when trying to use CSI Camera (from https://stackoverflow.com/questions/65631801/illegal-instructioncore-dumped-error-on-jetson-nano) edit /~/.bashrc file an put the following line to the end of the file:

export OPENBLAS_CORETYPE=ARMV8

Google Coral Install

From https://coral.ai/docs/m2/get-started/#2a-on-linux

First, add our Debian package repository to your system (be sure you have an internet connection):

echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list

curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -

sudo apt-get update

Then install the PCIe driver and Edge TPU runtime packages:

sudo apt-get install gasket-dkms libedgetpu1-std

Then install pycoral

sudo apt-get install python3-pycoral

install missing dependencies for TensorFlow

sudo pip3 install astor

sudo pip3 install Pillow

Test Coral

mkdir coral && cd coral git clone https://github.com/google-coral/pycoral.git cd pycoral bash examples/install_requirements.sh classify_image.py

python3 examples/classify_image.py
--model test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite
--labels test_data/inat_bird_labels.txt
--input test_data/parrot.jpg

Other stuff you might need

Install Jupyter Notbook

pip3 install --upgrade pip

pip3 install jupyter

start jupyter and create a ssh tunel to access remote via localhost:8000

jupyter notebook

ssh -L 8000:localhost:8888 [email protected]

Install bazel (version 4.1.0)

Install bazel if you need to build tools like mediapipe from source.

sudo nvpmodel -m0 && sudo jetson_clocks --fan # enabling max performance mode

cd ~ && mkdir bazel && cd bazel && wget https://github.com/bazelbuild/bazel/releases/download/4.1.0/bazel-4.1.0-dist.zip

sudo apt-get install build-essential openjdk-8-jdk zip unzip

unzip bazel-4.1.0-dist.zip

On Jetson Nano (2GB) you need to increase the JVM Memory before running compile.sh

export BAZEL_JAVAC_OPTS="-J-Xms1536m -J-Xmx1536m"

Run compile

env EXTRA_BAZEL_ARGS="--host_javabase=@local_jdk//:jdk" bash ./compile.sh

sudo cp ~/bazel/output/bazel /usr/local/bin/