Skip to content

Latest commit

 

History

History

movidius

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
measurements
measurements
 
 
model/v1
model/v1
 
 
Makefile
Makefile
 
 
README.md
README.md
 
 
image-classifier.py
image-classifier.py
 
 

README.md

Intel Movidius and Raspberry Pi

This folder contains the files required to run Inception v(1-4) on the Intel® Movidius™ Neural Compute Stick on a Raspberry Pi.

Setting up the environment

Note: The instructions below are heavily based on this sample app.

To generate the files for inference, follow the steps in Compiling the model. For those of you who just want to run inference on the stick itself, some precompiled files have been uploaded to the model directory. If this is the case check the Running inference section.

Compiling the model

NOTE: Execute the following instructions on a an x86_64 machine running Ubuntu 16.04 (development machine).

  1. Follow the official guide to install NCSDK v2 on the development machine.
  2. Get the sources:
mkdir -p ~/workspace/tensorflow

# Clone TensorFlow source and models repo
cd ~/workspace/tensorflow
git clone https://github.com/tensorflow/tensorflow --branch v1.7.1
git clone https://github.com/tensorflow/models

# Clone this repo
cd ~/workspace
git clone https://github.com/crespum/oscw18-edge-ai.git
  1. Compile a model:
# Download, export, freeze and profile model
cd ~/workspace/oscw18-edge-ai/movidius/
export TF_SRC_PATH=~/workspace/tensorflow/tensorflow
export TF_MODELS_PATH=~/workspace/tensorflow/models
make VERSION=v1 compile

Running inference

NOTE: Execute the following instructions on a Raspberry Pi with an Intel Movidius Neural Compute Stick.

  1. Install NCSDK v2 API. This will not install the entired SDK but just the interface for using the stick.
# Install dependencies
sudo apt-get install -y libusb-1.0-0-dev libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler libatlas-base-dev git automake byacc lsb-release cmake libgflags-dev libgoogle-glog-dev liblmdb-dev swig3.0 graphviz libxslt-dev libxml2-dev gfortran python3-dev python-pip python3-pip python3-setuptools python3-markdown python3-pillow python3-yaml python3-pygraphviz python3-h5py python3-nose python3-lxml python3-matplotlib python3-numpy python3-protobuf python3-dateutil python3-skimage python3-scipy python3-six python3-networkx python3-tk

# Compile and install API
git clone https://github.com/movidius/ncsdk -b ncsdk2
cd ncsdk/api/src
make
sudo make install
  1. Clone this repo
git clone https://github.com/crespum/oscw18-edge-ai.git
cd oscw18-edge-ai/movidius
make infer # See available parameters below

We can use two parameters:

  • VERSION to select the Inception version to compile (v1-v4)
  • INPUT to select on which we want to run inference. A sample image has been uploade to the repo too. We could for instance use make VERSION=v4 INPUT=foo.jpg

The output should look like this:

$ make infer
Running inferences using image-classifier project...
(python3 image-classifier.py --graph ./model/v1/graph_ncgraph --dim 224 224 --mean 127.5 --scale 0.00789 --colormode "RGB" --labels ./model/v1/labels.txt --image data/pic_002.jpg)

==============================================================
Top predictions for pic_002.jpg
Execution time: 88.5288ms
--------------------------------------------------------------
74.7%	288:lynx, catamount
8.1%	283:tiger cat
4.8%	282:tabby, tabby cat
4.1%	284:Persian cat
1.2%	286:Egyptian cat
==============================================================

If you want to use the files compiled in the previous section you need to copy these two files from the development machine to the Raspberry Pi (replace the name of the last folder accordingly):

cd ~/workspace/oscw18-edge-ai/movidius/
scp model/v1/labels.txt pi@raspberrypi.local:/home/pi/oscw18-edge-ai/movidius/model/v1/
scp model/v1/labels.txt pi@raspberrypi.local:/home/pi/oscw18-edge-ai/movidius/model/v1/