Powering Edge AI with the Powerful Jetson Nano

NVidia Jetson Nano Deep Learning Edge Device


Nano The Cat





 Hardware:
Jetson Nano developer kit. Built around a 128-core Maxwell GPU and quad-core ARM A57 CPU running at 1.43 GHz and coupled with 4GB of LPDDR4 memory! This is power at the edge. I now have a favorite new device.

 You need to add some kind of USB WiFi adaptor if you are not hardwired to ethernet. This is cheap and easy, I added a tiny $15 WiFi adapter and was off to the races.


Operating System:
Ubuntu 18.04

Library Setup:
sudo apt-get update -y
sudo apt-get install git cmake -y
sudo apt-get install libatlas-base-dev gfortran -y
sudo apt-get install libhdf5-serial-dev hdf5-tools -y

sudo apt-get install python3-dev -y
sudo apt-get install libcv-dev libopencv-dev -y
sudo apt-get install fswebcam -y
sudo apt-get install libv4l-dev -y
sudo apt-get install python-opencv -y
pip3 install psutil
pip2 install psutil
pip3.6 install easydict -U
pip3.6 install scikit-learn -U
pip3.6 install opencv-python -U --user
pip3.6 install numpy -U
pip3.6 install mxnet -U
pip3.6 install mxnet-mkl -U
pip3.6 install gluoncv --upgrade
sudo apt-get install libhdf5-serial-dev hdf5-tools libhdf5-dev zlib1g-dev zip libjpeg8-dev -y
sudo apt-get install python3-pip
sudo pip3 install -U pip
sudo pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v42 tensorflow-gpu
sudo nvpmodel -q --verbose
pip3 install numpy
pip3 install keras
git clone https://github.com/dusty-nv/jetson-inference
cd jetson-inference
git submodule update --init
tegrastats
pip3 install -U jetson-stats

 Source:
https://github.com/tspannhw/iot-device-install
https://github.com/tspannhw/minifi-jetson-nano

 IoT Setup

Download MiNiFi 0.6.0 Source from Cloudera and Build.
Download MiNiFi Java Agent (Binary)  and Unzip.

Follow these instructions.

On a Server

We want to hookup to EFM to make flow development, deploy, management and monitoring of MiNiFi agents trivial.   Download NiFi Registry.    You will also need Apache NiFi.

For a good walkthrough and hands-on demonstration see this workshop.

See these cool Jetson Nano Projects:  https://developer.nvidia.com/embedded/community/jetson-projects

Monitor Status
https://github.com/rbonghi/jetson_stats

Example Flow

 It's easy to add MiNiFi Java or CPP Agents to the Jetson Nano.   I did a custom NiFi CPP 0.6.0 build for Jetson.  I did a quick flow to run the jetson-inference imagenet-console CPP binary on an image captured from a compatible Logitech USB Webcam with fswebcam.   I store the images to /opt/demo/images and pass it on the command line to the CPP console as a proof of concept.

#!/bin/bash

DATE=$(date +"%Y-%m-%d_%H%M")

fswebcam -q -r 1280x720 --no-banner /opt/demo/images/$DATE.jpg

/opt/demo/jetson-inference/build/aarch64/bin/imagenet-console  /opt/demo/images/$DATE.jpg  /opt/demo/images/out_$DATE.jpg
==
imagenet-console
  args (3):  0 [/opt/demo/jetson-inference/build/aarch64/bin/imagenet-console]  1 [/opt/demo/images/2019-07-01_1405.jpg]  2 [/opt/demo/images/out_2019-07-01_1405.jpg]


imageNet -- loading classification network model from:
         -- prototxt     networks/googlenet.prototxt
         -- model        networks/bvlc_googlenet.caffemodel
         -- class_labels networks/ilsvrc12_synset_words.txt
         -- input_blob   'data'
         -- output_blob  'prob'
         -- batch_size   2

[TRT]  TensorRT version 5.0.6
[TRT]  detected model format - caffe  (extension '.caffemodel')
[TRT]  desired precision specified for GPU: FASTEST
[TRT]  requested fasted precision for device GPU without providing valid calibrator, disabling INT8
[TRT]  native precisions detected for GPU:  FP32, FP16
[TRT]  selecting fastest native precision for GPU:  FP16
[TRT]  attempting to open engine cache file /opt/demo/jetson-inference/build/aarch64/bin/networks/bvlc_googlenet.caffemodel.2.1.GPU.FP16.engine
[TRT]  loading network profile from engine cache... /opt/demo/jetson-inference/build/aarch64/bin/networks/bvlc_googlenet.caffemodel.2.1.GPU.FP16.engine
[TRT]  device GPU, /opt/demo/jetson-inference/build/aarch64/bin/networks/bvlc_googlenet.caffemodel loaded
[TRT]  device GPU, CUDA engine context initialized with 2 bindings
[TRT]  binding -- index   0
               -- name    'data'
               -- type    FP32
               -- in/out  INPUT
               -- # dims  3
               -- dim #0  3 (CHANNEL)
               -- dim #1  224 (SPATIAL)
               -- dim #2  224 (SPATIAL)
[TRT]  binding -- index   1
               -- name    'prob'
               -- type    FP32
               -- in/out  OUTPUT
               -- # dims  3
               -- dim #0  1000 (CHANNEL)
               -- dim #1  1 (SPATIAL)
               -- dim #2  1 (SPATIAL)
[TRT]  binding to input 0 data  binding index:  0
[TRT]  binding to input 0 data  dims (b=2 c=3 h=224 w=224) size=1204224
[cuda]  cudaAllocMapped 1204224 bytes, CPU 0x100e30000 GPU 0x100e30000
[TRT]  binding to output 0 prob  binding index:  1
[TRT]  binding to output 0 prob  dims (b=2 c=1000 h=1 w=1) size=8000
[cuda]  cudaAllocMapped 8000 bytes, CPU 0x100f60000 GPU 0x100f60000
device GPU, /opt/demo/jetson-inference/build/aarch64/bin/networks/bvlc_googlenet.caffemodel initialized.
[TRT]  networks/bvlc_googlenet.caffemodel loaded
imageNet -- loaded 1000 class info entries
networks/bvlc_googlenet.caffemodel initialized.





Reference:




on