OpenCV Face Detection

These are my notes based on Jason Dsouza’s excellent OpenCV video course.

Installation

python3 -m pip install opencv-contrib-python

Finding a Classifier

OpenCV’s face detection loads a cascade classifier which is then used to detect faces.

Go to github.com/opencv/opencv/data/haarcascades. There are a number of different classifiers, download haarcascade_frontalface_default.xml

Running the classifier

Using an image with a face (or many faces); we can detect them with:

import cv2 as cv

img = cv.imread('Photos/people.jpg')

cascade_file = 'haarcascade_frontalface_default.xml'
haar_cascade = cv.CascadeClassifier(cascade_file)
faces_rect = haar_cascade.detectMultiScale(
  cv.cvtColor(img, cv.COLOR_BGR2GRAY), 
  scaleFactor=1.1, minNeighbors=5)

for (x,y,w,h) in faces_rect:
    cv.rectangle(img, (x,y), (x+w, y+h), (0,255,255), 2)
cv.imshow('Detected faces', img)

cv.waitKey(0)

For the Super Mario Movie cast, I got these results.

There were a few false positives and negatives.

This could be further tuned by adjusting the scaleFactor and minNeighbours parameters; however, this isn’t the best idea since we would only be tuning the face detection to work with a specific type of image.