AI Vision Academy

Access the code of this tutorial, computer vision courses and an exclusive community on AI Vision Academy

• Access to over 50+ source codes from Pysource.com/blog
• Dedicated video courses about computer vision
• Access to an exclusive community of professionals
• Real-World AI Projects – Get hands-on experience building practical AI Computer Vision solutions with a structured path.
• Monthly Coaching Calls – get support and any of your questions answered

Subscribe to our newsletter to learn more

❌

What is face recognition?

With face recognition, we not only identify the person by drawing a box on his face but we also know how to give a precise name. With OpenCV and Python, through a database, we compare the person’s photo and we know how to identify it precisely.

In this tutorial, we will use as a basis a code created by Adam Geitgey and this is the original GitHub Project repository of Github, if you want you can easily see all the details and the potential. In this lesson, we will make everything simple and easy to follow but as always I invite you to develop the project starting from what you will learn today.

Here’s what you will find:

1. Installations
2. Face recognition on image
3. Face recognition in real-time on a webcam
4. Let’s do some tests with facial recognition

1. Installations

For convenience, I invite you to download the package with all the codes and photos that you will find in my lesson, and then we proceed with the installation of the basic libraries.

The first library to install is opencv-python, as always run the command from the terminal.

pip install opencv-python

then proceed with face_recognition, this too installs with pip.

pip install face_recognition

2. Face recognition on image

To make face recognition work, we need to have a dataset of photos also composed of a single image per character and comparison photo. For example, in our example, we have a dataset consisting of 1 photo each of Elon Musk, Jeff Bezos, Lionel Messi, Ryan Reynolds, Sergio Canu.

and in the comparison, we will use the photo of Messi

Call the libraries

The first step is always to recall the libraries we have installed OpenCV and face_recognition in our project.

import cv2
import face_recognition

Face encoding first image

With the usual OpenCV procedure, we extract the image, in this case, Messi1.webp, and convert it into RGB color format. Then we do the “face encoding” with the functions of the Face recognition library.

img = cv2.imread("Messi1.webp")
rgb_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
img_encoding = face_recognition.face_encodings(rgb_img)[0]

Face encoding second image

Same procedure for the second image, we only change the name of the variables and obviously the path of the second image, in this case: images/Messi.webp.

img2 = cv2.imread("images/Messi.webp")
rgb_img2 = cv2.cvtColor(img2, cv2.COLOR_BGR2RGB)
img_encoding2 = face_recognition.face_encodings(rgb_img2)[0]

Comparison of images

With a single line, we make a simple face comparison and print the result. If the images are the same it will print True otherwise False.

result = face_recognition.compare_faces([img_encoding], img_encoding2)
print("Result: ", result)

Encode all face in the dataset

Now we have to encode all the images in our database, so that through the webcam video stream if it finds the match it shows the name otherwise it says “name not found”.

This is a function of the file I have prepared and it simply takes all the images contained in the images/ folder and encodes them. In our case, there are 5 images.

# Encode faces from a folder
sfr = SimpleFacerec()
sfr.load_encoding_images("images/")

3. Face recognition in real-time on a webcam

Even for face recognition in real-time, the procedure is similar to that of a single image but with something more. As a first step remember to download the files from the link below and among the various Python files you will also find simple_facerec.py, remember that this is not a library so to include it in your project, put this file in the same folder and these are the correct lines of code to start.

import cv2
from simple_facerec import SimpleFacerec

Take webcam stream

With a simple Opencv function, We take the webcam stream and loop it

# Load Camera
cap = cv2.VideoCapture(2)


while True:
    ret, frame = cap.read()

Face location and face recognition

Now we identify the face passing the frame of the webcam to this function detect_known_faces(frame). It will give us the name of the person and an array with the position at each moment of the movement.

    # Detect Faces
    face_locations, face_names = sfr.detect_known_faces(frame)
    for face_loc, name in zip(face_locations, face_names):
        y1, x2, y2, x1 = face_loc[0], face_loc[1], face_loc[2], face_loc[3]

As an intermediate step, I have shown the output in the image.

Show name and rectangle

Now that we have all the elements we show them with OpenCV. For more details on this part, I suggest you also see How crop images with OpenCV and Python

        cv2.putText(frame, name,(x1, y1 - 10), cv2.FONT_HERSHEY_DUPLEX, 1, (0, 0, 200), 2)
        cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 200), 4)

    cv2.imshow("Frame", frame)

    key = cv2.waitKey(1)
    if key == 27:
        break

Here’s how it shows me and my name

4. Let’s do some tests with facial recognition

For the real-time test, I put the photos on my phone and showed them to the webcam. Even if the lighting conditions are not perfect it works quite well and obviously, in real conditions, it will work much better.