In 2001, Paul Viola of Microsoft Research and Michael Jones of Mitsubishi Electric developed a revolutionary method of detecting faces in an image by developing a classifier called the Haar cascade classifier https://www.face-rec.org/algorithms/Boosting-Ensemble/16981346.pdf. The Haar cascade classifier is based on Haar features, which are the sum of the difference of pixel values in a rectangular region. The magnitude of the difference value is calibrated to indicate the characteristics of a given region in the face—for example, nose, eyes, and so on. The final detector has 38 cascade classifiers with 6,060 features consisting of about 4,916 face images and 9,500 non-face images. The total training time was several months, but the detection time was very fast.
First, the image...
