About movidius: https://developer.movidius.com/

(Skip caffe compared with offical document https://developer.movidius.com/start)

1. Install vim and opencv:

sudo apt-get install vim
sudo apt-get install python-opencv

2. Create workspace folder:

cd /home/pi
mkdir -p workspace
cd workspace

3. Clone ncsdk from movidius:

git clone https://github.com/movidius/ncsdk.git
cd ncsdk

4. Use vim or other editor to remove line "./install-ncsdk.sh" in file install.sh (about line 40).

5. Run install script:

./install.sh

6. Change folder:

cd /opt/movidius/NCSDK

7. Use vim or other editor to remove line "check_and_install_caffe" in file install-ncsdk.sh (about line 535).

8. Run script (takes about thirty minutes):

./install-ncsdk.sh

9. Run example:

cd /home/pi/workspace/ncsdk/examples/tensorflow/inception_v1
python run.py

1. Use "sudo raspi-config" choose "interface option -> camera" to enable camera and then reboot
2. Clone our repo:

cd /home/pi/workspace 
git clone https://github.com/ihere1/movidius-face

3. Run (Enter 'q' in video window to exit program):

cd movidius-face/face_detect_pi/
python pi_two_devices.py

Follow https://developer.movidius.com/start install packages needed on your Ubuntu and RPI3.

1. On Windows / Ubuntu machine with tensorflow, you can run python run_model.py to process pictures in run_model_input folder. Press any key to switch picture.
2. On Linux with two sticks, you can run pi_two_devices.py to process pictures in two_devices_input folder. Press any key to switch picture.

Files in face_detect_pi folder:

Commands on Ubuntu to generate checkpoint folders and graph files: