Facial Recognition on Raspberry Pi 2025

Materials: Raspberry Pi 4 or 5 and Webcam

This project walks you through setting up facial recognition on a Raspberry Pi using a virtual environment, OpenCV, and the face-recognition library. With the release of Bookworm OS, virtual environments are now required to avoid conflicts with system packages — but don’t worry, it’s quick and easy!

Watch the Tutorial on YouTube - Face Rec on RPi

Set Up Virtual Environment and Install Libraries

Open a terminal window and follow these steps:

Create a virtual environment

python3 -m venv --system-site-packages face_rec

Activate the virtual environment

source face_rec/bin/activate

Update your Raspberry Pi:

sudo apt update && sudo apt full-upgrade -y

Add Swap Memory (Recommended First Step)

The Raspberry Pi doesn’t have enough memory to compile dlib. Add swap space to avoid memory crashes during compilation:

Open the dphys-swapfile config

sudo nano /etc/dphys-swapfile

Find the line:

CONF_SWAPSIZE=512

Change it to:

CONF_SWAPSIZE=2048

Save and exit (CTRL+X, then Y, then Enter), then restart the swap service:

sudo systemctl restart dphys-swapfile

Install Required Python Libraries

pip install opencv-python

pip install imutils

Install CMake (for compiling dependencies):

sudo apt install cmake -y

Installing face recognition will take anywhere from 10 minutes to an hour.

pip install face-recognition

Once you're done with installation, it's a good idea to change the swap size back to reduce SD card wear:

sudo nano /etc/dphys-swapfile

Change:

CONF_SWAPSIZE=2048

Back to:

CONF_SWAPSIZE=512

Then restart the service:

sudo systemctl restart dphys-swapfile

If you ever exit the terminal, simply re-run source face_rec/bin/activate to reactivate your environment.

Download the Code

Clone this repository:

git clone https://github.com/carolinedunn/facial_recognition.git

Change into the directory:

cd facial_recognition

Delete the sample directory:

rm -r dataset/Z

Take Headshots

If using a webcam, first modify the code with the person’s name in the dataset directory:

nano headshots_capture-webcam.py

If using a Pi Camera:

nano headshots_capture-picam.py

Change line 7 of the file replacing YOUR_NAME:

PERSON_NAME = "YOUR_NAME"

Save and exit by pressing Ctrl+X, then Y, and hit Enter.

Now run the script for the webcam:

python3 headshots_capture-webcam.py

Here's the command if using a Pi Camera:

python3 headshots_capture-picam.py

Look at the camera and press the spacebar to take photos. Move your head around and take at least 10 photos.

Press q to exit. Repeat this for each person.

You should now see a folder for each person with a set of headshots.

Train the Model

python3 model_training.py

If successful, you will get a .pickle file.

Run the Facial Recognition Test

If using a webcam:

python3 face_rec-webcam.py

If using a Pi Camera:

python3 face_rec-picam.py

Press q to exit.

Original Tutorial - https://www.tomshardware.com/how-to/raspberry-pi-facial-recognition
Updated Code from Core Electronics - https://core-electronics.com.au/guides/face-recognition-with-raspberry-pi-and-opencv/