Pan/tilt face tracking with a Raspberry Pi + NCS

This project using the NCS with openvino and ServoBlaster to drive multiple servos via the GPIO pins to face tracking

Demo [YouTube] [Bilibili]

Installation Python Libraries:

• Python 3.5
• Picamera
• OpenVINO
• Numpy
• OpenCV

Things needed:

• A raspberry pi 3B
• A Intel® Neural Compute Stick - (NCS1/2)
• A pan/tilt bracket - (3D printer)
• Two Servos - (SG90)
• A GPIO expansion board
• Pi Camera or USB Webcam

Pan/Tilt bracket

Pan-and-tilt bracket - (3D printer)

[GPIO 4 -> PanMotor] [GPIO 17 -> TiltMotor]

The code defaults to driving 8 servos, the control signals of which should be connected to P1 header pins as follows:

Servo number    GPIO number   Pin in P1 header   Pan-Tilt Motor
      0               4             P1-7           Pan-Motor
      1              17             P1-11          Tilt-Motor
      2              18             P1-12
      3             21/27           P1-13
      4              22             P1-15
      5              23             P1-16
      6              24             P1-18
      7              25             P1-22

Install the OpenVINO™ Toolkit for Raspbian* OS Package

FaceDetection model (IR) ./models:

• Network - face-detection-retail-0004.bin
• Weights - face-detection-retail-0004.xml

Face detector based on SqueezeNet light (half-channels) as a backbone with a single SSD for indoor/outdoor scenes shot by a front-facing camera.

METHOD 1:

• Run this script ./Install_openvino.sh

This script provides all instructions on install the OpenVINO™ toolkit package for Raspbian OS*

METHOD 2:

• The following steps will be covered: [Guide]

This guide provides step-by-step instructions on how to install the OpenVINO™ toolkit for Raspbian OS*

To test your OpenVINO, open a new terminal. You will see the following:

[setupvars.sh] OpenVINO environment initialized

Getting Started:

Install and start multiple servos:

git clone [email protected]:yehengchen/FaceTracking-RPI3-NCS.git
cd FaceTracking-RPI3-NCS/ServoBlaster/user
sudo ./servod

Multiple servos testing:

echo 0=+10 > /dev/servoblaster
echo 1=+10 > /dev/servoblaster

Picamera testing:

raspistill -o image.jpg

Run face tracking:

python3 pi_NCS_face_traking.py

Reference:

PiBits-ServoBlaster