In this tutorial, we will learn how to detect eyes from an image using the Haar Cascades classifier. For the well-known tasks, the classifiers/detectors already exist, for example: detecting things like faces, cars, smiles, eyes, and license plates.

Object Detection using Haar feature-based cascade classifiers is an effective object detection approach recommended by Paul Viola and Michael Jones in their paper, “Rapid Object Detection using a Boosted Cascade of Simple Features” in 2001.

Installing the modules

To install the OpenCV module and some other associated dependencies, we can use the pip command:

pip install opencv-python

pip install numpy

Download the Haar Cascades

1. Follow the URL:
   https://github.com/opencv/opencv/tree/master/data/haarcascades/
2. Click on haarcascade_eye.xml
3. Click on Raw and then press Ctrl + S. This will help you save the Haar Cascade file for eyes.

Source code:

import cv2
import numpy as np

cascade = cv2.CascadeClassifier('haarcascade_eye.xml')

org_img = cv2.imread('kid.jpg') # reading image 
img     = cv2.resize(org_img, (500,500)) # resizing image size
copy    = img.copy() # copying the image
gray    = cv2.cvtColor(copy, cv2.COLOR_BGR2GRAY) # convert image

eyes    = cascade.detectMultiScale(gray, 1.3, 5)
for (ex, ey, ew, eh) in eyes:
	cv2.rectangle(copy, (ex, ey), (ex+ew, ey+eh), (155, 155, 0), 2)

stack = np.hstack([img, copy])
cv2.imshow('Output', stack)
cv2.waitKey(0)

The Output