main.py

import cv2

# initiates video capture
video = cv2.VideoCapture(0) #Sets 0 at camera ID
video.set (3, 640) # set width
video.set (4, 480) # set height

# Add haar cascade classifiers
#Data of frontal face classifiers
frontal_cascade = cv2.CascadeClassifier(cv2.data.haarcascades 
                                        + 'haarcascade_frontalface_default.xml')

#Data of profile face classifiers
profile_cascade = cv2.CascadeClassifier(cv2.data.haarcascades 
                                        + 'haarcascade_profileface.xml')

# Loop that initiates the camera
while True:
    capture, img=video.read()

    # Converts image into grayscale  
    img_gray=cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    # Detects faces on image
    face_frontal = frontal_cascade.detectMultiScale(
        img_gray, 
        scaleFactor=1.3, # Reduces the image
        minNeighbors=4   # Defines the number rectanglesnto detect a face
    )

    face_profile = profile_cascade.detectMultiScale(
        img_gray, 
        scaleFactor=1.3, # Reduces the image
        minNeighbors=4  # Defines the number rectangles to detect a face
    )
                                                                             
    # Join both classifiers
    merge_faces=list(face_frontal) + list (face_profile)

    # Aplied gaussian blurr for every face detected
    for (x, y, w, h) in merge_faces:

        # Defines the submatrix of the region we want to blurr
        roi = img[y:y + h, x:x + w]

        # Applies Gaussian blur to the "roi"
        blur = cv2.GaussianBlur(roi, (99, 99), 0)

        # Replace the original face region with the blurr
        img[y:y + h, x: x + w] = blur

    #Open camera
    cv2.imshow("Face blurring", img) # Pass image frame 

    # Standart quote to close de camera
    if cv2.waitKey(1) & 0xFF==ord('q'): # Press 'q' from the keyboard to close the program 
        break

#Release the window resources at close
video.release()
cv2.destroyAllWindows()