Face Detection using HAAR Cascade Step by Step In the paper, the author makes use of Ada Boost as it gives the best results.

We however need to make sure that have a good mix of images with faces and images without faces to train the classifier in a robust manner. We can use thousands of labeled images convert them into the HAAR cascade feature map and train it using a machine learning model for creating a face detection classifier. However, calculating the sum of pixels in white and black regions throughout the images can be a costly operation especially for large images, so the authors also proposed a method known as an integral image that can achieve the same calculation by performing the operations on just four pixels. So using these rectangular features over an image we can produce thousands of feature points for the image.

And the second rectangular feature is measuring the difference in intensity between the two eye regions and the nose bridge. In the below illustration, the first rectangular feature is calculating the difference in intensity between the eye regions and cheeks regions of the face. The rectangular features are slid across the image and the sum of the pixel that lies in the white part is subtracted from the sum of the pixels falling in the black part. The rectangular features which are similar to the kernel are used to detect different features of the face like eyes and notes as shown in the illustration. In their paper, the author presented a new and faster method of processing images and detecting faces using rectangular features as shown below. However, there pre-trained HAAR cascade available in their GitHub for other objects as well like for full body, upper body, lower body, smile, and many more. The original implementation is used to detect the frontal face and its features like Eyes, Nose, and Mouth. HAAR cascade is a feature-based algorithm for object detection that was proposed in 2001 by Paul Viola and Michael Jones in their paper, “Rapid Object Detection using a Boosted Cascade of Simple Features”. We will explain to you how HAAR Cascade works and then show you how to perform face detection with HAAR Cascade in OpenCV. It is quite popular in the computer vision community for being excellent object detectors and classifiers due to the fact that they are lightweight, easy to use, and have a very high inference speed. In this article, we will be taking a thorough look into the HAAR cascade family of classifiers.

3.1 Face Detection using HAAR Cascade Step by Step.

Face detection opencv