SAS takes attendance using facial recognition, intended to replace paper-based traditional attendance sheets. This can be seen as a practical usage of Amazon Web Services such as rekognition, S3, and Polly.

• 1 raspberry pi (we used rPi 3 model B)
• 1 microUSB to USB power cord for rPi
• 1 USB webcam
• 3 LED lights (one red, one yellow, one green)
• 1 small breadboard (see image)
• 8 male-female connectors
• 1 button
• 1 mobile phone battery

You must run the below commands on the raspberry Pi:

sudo apt-get install boto
sudo apt-get install boto3
sudo apt-get install fswebcam
sudo apt-get install python-pygame

Also consider turning up the volume on your raspberry pi. T here is a nice tool to do this called alsamixer.

Have the female sides of the connectors connected to the raspberry pi. For each LED and for the button, there must be one connector connected to a GND (ground) pin, and one connected to a GPIO (general purpose input/output) pin. The code here uses BCM numbering, so the LEDs and button are numbered by GPIO port. For example, and LED connected to GPIO17 would be numbered as 17.

The configuration of the breadboard is shown below:

For each light, there is a GPIO pin connected to the jack going directly to the LED. There is a resistor going between the LED and the other jack, which leads to a GND pin. Although it may not be entirely clear, here are some images of where the connectors are inserted into the rPi:

When on standby, only the red light is on. After the user presses the button, the yellow light starts blinking and a photo is taken. The yellow light keeps on blinking until the API call to AWS Rekognition is successful and the name of the user is returned.

You need Git bash or an equivalent to do the below commands.

Copy all files in laptop folder to pi folder:

scp -P 2222 * [email protected]:foldername

Connect to rPi:

ssh [email protected] -p 2222

The Pi that was configured for this project is listening for SSH on port 2222 for security purposes. Remove the -p 2222 on the commands if you are using a fresh raspberry pi. Change the ip address with the ip address of your pi.

Note that this system is viable. The current runtime can be more than halved if:

1. The USB webcam is swapped for an official camera module
2. AWS Polly is replaced for a local text to speech program or an LCD display.
3. A stable wifi connection is obtained or the system is hooked up to ethernet cable

There are several ways that you can improve this project.

• Get the official raspberry pi camera module. Most webcams are not reliable and require you to stay still for some time to take a photo.
• Instead of calling AWS polly, consider using a built-in text to speech library. This can dramatically improve the runtime.
• Get a bluetooth speaker or a micro speaker which can be embedded inside the case. As an alternative, replace that entirely with an LCD display.
• Implement automatic photo-taking when face detected instead of having the user manually pressing the photo-taking button. This can be done with openCV.
• Replace the mobile phone charging battery for a long rPi power cord that plugs into a wall socket.
• Connect the pi’s attendance sheet to your computer via network shared folder so that it automatically synchronizes.
• Ensure that you have a reliable internet connection or connect the system to LAN.
• Write a script or find some method to quickly and conveniently upload new faces and names to the AWS Rekognition Collection.

Watch our demo video of the final prototype here!

Joon Park

Andrew Luo