Team

Name	Responsibility	GUID
Oscar Meunier	Network and Design engineer	2391076M
Sai Vamsi Karnam	Development and Integration engineer	2805927K
Tara Duggan	Electronics and Design engineer	2363956D
Rohith Kambampati	Backend developer and programmer	2782679K

Track-Record is a face tracking system built on the Raspberry Pi 3B platform, using a Pi Camera Module and servo motors for real-time face detection and tracking. A GUI is also implemented which communicates via a server sending commands to the Raspberry Pi.

The full documentation can be found here.

Find us on Twitter and Instagram to stay up to date with any of the latest developements or email us at tracknrecordteam@gmail.com to get in touch.

Hardware

The following electronic components were used to build the project:

• Raspberry Pi 3B

• [SD Card] - One with a memory capacity of 16 gigs or more.

• Pi Camera module 2

• SG92R Servo

Software Dependencies

The following libraries were used throughout the project:

• CMake
• OpenCV
• BCM2835
• QT5

Installation and Usage

This project aims to build a standalone plug and play type product. Clip the camera of your choice to the platform, switch the power on and watch it track you. But wait there's more, you can build it on your own. Follow the instructions ahead.

Instructions

• Step 1: Setting up the hardware

  • Install the OS and set up a display.*
  • Connect the Pi camera
  • Connect the motors using the PinOut diagram: Pan at Gpio 18, Tilt at Gpio 17
  • 3D print the models -- Check the Schematics and Solidworks directories or build your own model

• Step 2: Software

  • Make sure to have the latest update for the GCC/G++ compiler on your Raspberry Pi OS.
  • Setup cmake on your Raspberry Pi.
  • Clone the repo, the libraries used are included with the git package.

  git clone https://github.com/oscell/Track-Record.git
  

  • Unpack the lib_package archive to /usr/include

  sudo tar -xf lib_package.tar.gz -C /usr/include

  • Create an executable in the main/build directory

    • Open a terminal and navigate to the main directory
    • Run the following commands to create the executable "tracknrecord"

    mkdir -p build && cd build
    cmake ..
    make

    • You should now see two executables named "tracknrecord" and "server" in the directory

  • Set executable permission to run on startup

    • Open a terminal and run the following commands

    mkdir /home/pi/.config/autostart
    nano /home/pi/.config/autostart/track.desktop

    • Copy or type the following text into the "track.desktop" file.
    • Replace the "directory" with the location of the executable named 'server.cpp' located in the main folder. (eg. home/pi/...)

    [Desktop Entry]
    Type=Application
    Name=Track
    Exec="directory"

    • Save the file using: Ctrl+X -> Y

• Step 3:

  • Setup the GUI* as shown in section 2 of the footnotes.

• Step 4:

  • Reboot the Pi from the terminal and it is ready to go.

    sudo reboot

  • The Pi should now be listening for commands.

  • Ensure the Pi and your personal machine are on the same network.

  • Open the GUI on your personal machine and click "start tracking".

  • To view the video feed, click the video feed button.

---

*Footnotes

Section 1: Additonal information

• Raspberry Pi PinOut diagram
• SG92r wiring: Red for Power, Brown for Ground, and Yellow/Orange for Data.
• Set up Raspberry Pi in headless mode without a display and other hardware.

Section 2: GUI

This section should be built locally (not on the raspberry PI). It will be used to comunicate with the PI to send commands.

Dependencies

As previously with the raspi install cmake on your machine. The next step is to install Qt5 although this can be done simply on linux using the command:

sudo apt-get install qt5-default

To find the path of your Qt5 installation, set up the C++, cmake, and QT5 on your personal machine and run the following command in the terminal:

find / -iname Qt5Config.cmake 2>/dev/null

Once you have the path, you can update the /local/CMakeLists.txt file with the correct Qt5_DIR.

set(Qt5_DIR "/path/to/Qt5/lib/cmake/Qt5")

The comunication with the raspberry pi is done through a server. To find the IP adress of the raspery pi run in the raspberry pi terminal:

hostname -I

In /local/src/GUI/main.cpp change:

QString serverAddress = "127.0.0.1";

to

QString serverAddress = "<raspberry_pi_ip>"

Build

cd local/
mkdir -p build && cd build
cmake ..
make

Usage

On the Raspi

Before being able to run send command the server must be initialised on the raspi.

cd main/build/
./server 

On your local machine

Run the GUI.

 cd local/build/
./GUI1_0

It is comprised of a section for vieo feed input and 2 buttons.

Tracking button

This sends a signal to the client which toggles the tracknrecord executable depending on the state it is currently in.

Video feed

This toggles the user webcam to show how face detection and tracking works. The GUI should look like this:

Section 3: Testing

The test report and the code we used for tests is available on a different branch of this repo. It can be accessed using https://github.com/oscell/Track-Record/tree/tests/tests

Section 4: 3D printing material

The Solidworks assemblies are provided in Solidworks/. This will require version 2022 or above however. To print the latest version of the stand .stl files are available in Solidworks/Stls/.

Exploded view

Note: In out product demonstration the base used was the Base_old as we did not have time to 3D print such a large base.

Once assembled model looks like:

and in exploded view: