How to Auto Blur Faces in Photos and Videos Using OpenCV (with Face Recognition Toggle)

Whether you're working on a security project, anonymizing people in public footage, or building a social app with privacy in mind—face blurring is essential. In this post, we’ll walk you through building a face detection and blurring tool using OpenCV. Bonus? You can also toggle face recognition using the face_recognition library to exclude known faces from being blurred.

What You’ll Learn

• Detect faces using Haar Cascades (or optionally MTCNN for better accuracy)

• Blur detected faces for privacy using OpenCV

• Bonus: Add a face recognition toggle to protect known faces

Libraries You’ll Need

pip install opencv-python face_recognition

For MTCNN:

pip install mtcnn

Step 1: Detect and Blur Faces Using Haar Cascades

import cv2

# Load Haar cascade
face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')

def blur_faces(image):
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray, 1.3, 5)

    for (x, y, w, h) in faces:
        face = image[y:y+h, x:x+w]
        face = cv2.GaussianBlur(face, (99, 99), 30)
        image[y:y+h, x:x+w] = face

    return image

Exclude Known Faces from Blurring

import face_recognition
import numpy as np

# Load known face
known_image = face_recognition.load_image_file("your_face.jpg")
known_encoding = face_recognition.face_encodings(known_image)[0]

def blur_faces_with_recognition(image, skip_recognition=False):
    rgb_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    face_locations = face_recognition.face_locations(rgb_image)
    face_encodings = face_recognition.face_encodings(rgb_image, face_locations)

    for (top, right, bottom, left), encoding in zip(face_locations, face_encodings):
        if skip_recognition or not face_recognition.compare_faces([known_encoding], encoding)[0]:
            face = image[top:bottom, left:right]
            face = cv2.GaussianBlur(face, (99, 99), 30)
            image[top:bottom, left:right] = face

    return image

 

Testing the Code

# Load and blur
image = cv2.imread("group_photo.jpg")

# True = skip known face check, False = blur everyone
output = blur_faces_with_recognition(image, skip_recognition=False)

cv2.imshow('Blurred', output)
cv2.waitKey(0)
cv2.destroyAllWindows()

Using MTCNN Instead of Haar

from mtcnn.mtcnn import MTCNN

detector = MTCNN()

def blur_faces_mtcnn(image):
    rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    faces = detector.detect_faces(rgb)

    for face in faces:
        x, y, width, height = face['box']
        x, y = abs(x), abs(y)
        face_crop = image[y:y+height, x:x+width]
        face_crop = cv2.GaussianBlur(face_crop, (99, 99), 30)
        image[y:y+height, x:x+width] = face_crop

    return image

Use Cases

• Surveillance systems — blur passerby faces

• Public datasets — anonymize sensitive data

• Social apps — let users choose what gets shown

• Healthcare & Education — protect identities in media

Final Thoughts

Face detection and blurring is a powerful way to build privacy-first systems. Whether you're scrubbing video for compliance or just blurring photo-bombers in vacation photos, you now have a working solution.

Next step? Wrap it into a web or mobile UI, or integrate into your video pipeline with cv2.VideoCapture.