Follow face using OpenCV, Arduino and Webcam

Python script (and Arduino sketch) to follow face using OpenCV, 28BYJ-48 stepper & servo and Webcam.

This script + arduino with stepper motor as yaw rotation and servo motor as pitch for the webcam, will basically follow any visible face. There are two versions, one using haar cascades (faster but bad detection) and other one using Deep Neural Network & SSD (slower but precise).

Prerequirements

Windows

For windows install Python 3 and add it to your PATH. Next install pip3 and opencv-python, pyserial, imutils and numpy.

Ubuntu

Run these commands to install dependencies

apt install python3 python3-pip

pip3 install opencv-python pyserial imutils numpy

For Arduino IDE, all libraires used are already in IDE. (Servo.h & Stepper.h)

Usage

Arduino & motors

Firstly connect your ULN2003 driver board to power and these pins on the Arduino:

IN1 ->> D8
IN2 ->> D9
IN3 ->> D10
IN4 ->> D11

And also connect your servo motor to power and its signal cable to pin 6 on the Arduino.

Ensure that if you move the servo to 0°, the webcam is looking few degrees down and not hiting anything.

Now you can upload the code in 'ardcode' folder to your Arduino.

Python script

Open the one of the folders in command prompt or terminal.

Now you can start the python script using followface.py on Windows or python3 followface.py on Linux and enter the values of the serial port of the Arduino, webcam number and if you want to see detected faces etc.

OR

You can enter config file after it and skip all of that like this: followface.py example.txt or python3 followface.py example.txt on linux.

In the DNN version, you can press 'e', now the window frame will be updated every 30 frames, increasing the FPS by 2 frames (at least for me)

To exit, just press 'q'.

BotControl.exe

This app works only on Windows and you need to have .NET framework installed (https://dotnet.microsoft.com/download/dotnet-framework) (it can also work on linux using Wine, idunno)

Start the app, select port, click connect and now you can control the Arduino using your arrows.

Pictures/Videos

tobedone

Credits & Copyright

DNN is code from https://github.com/sr6033/face-detection-with-OpenCV-and-DNN that I modified. Haar code is done by me using references and tutorials. Arduino code is done by me. BotControl is done by me.

Everyone is allowed to use this code & modify it / publish it modified. If you do so, credit would be cool! :D