OpenVino docker and application example

These steps will install openvino in a docker container and will run a squeezenet demo on an Intel CPU and on a NCS.

Install docker-ce

curl -fsSL https://get.docker.com -o get-docker.sh
sh get-docker.sh
sudo usermod -aG docker $USER
# reload user groups
exec su -l $USER

Build the docker image

Clone this repository and run:

./build.sh

Add the USB access rights

cat <<EOF > 97-usbboot.rules
SUBSYSTEM=="usb", ATTRS{idProduct}=="2150", ATTRS{idVendor}=="03e7", GROUP="users", MODE="0666", ENV{ID_MM_DEVICE_IGNORE}="1"
SUBSYSTEM=="usb", ATTRS{idProduct}=="2485", ATTRS{idVendor}=="03e7", GROUP="users", MODE="0666", ENV{ID_MM_DEVICE_IGNORE}="1"
SUBSYSTEM=="usb", ATTRS{idProduct}=="f63b", ATTRS{idVendor}=="03e7", GROUP="users", MODE="0666", ENV{ID_MM_DEVICE_IGNORE}="1"
EOF
sudo mv 97-usbboot.rules /etc/udev/rules.d/
sudo udevadm control --reload-rules
sudo udevadm trigger
sudo ldconfig

Test OpenVino on an Intel CPU

./demo_cpu.sh

Test OpenVino on a NCS

./demo_myriad.sh

Training our own classifier and running the inference on the Neural compute stick

Here, we will see how to train a custom image classifier and run it on the NCS.

Setup fast.ai

See fastai-docker

Potatoes/Tomatoes dataset

The potatoestomatoes dataset can be downloaded from here.

Training

Train the network and export it to Onnx using the notebook: lesson1_export_to_onnx-pomatoes.ipynb

To get the notbook working, copy it to the fastai/courses/dl1 folder.

Inference

Model

Copy the Onnx model to code/pomatoes/Modelpotatoestomatoes_softmax.onnx in this repository

Labels

Edit the file named code/pomatoes/pomatoes.labels, checking that each line contains the correct category.

Preprocessing

Edit the file named code/pomatoes/classify_pomatoes.py, verify that the variables resnet_mean and resnet_scale have the correct values

Run

Plug in an NCS, and run:

./run_with_usb.sh pomatoes/pomatoes.sh

This script will first convert the model to run on the NCS with OpenVino and then run the inference on the images located in code/pomatoes/tomatoes_potatoes_images