Implementation and evaluation of K-nearest neighbors (KNN) algorithm for human face recognition

Authors: Maria Chiritescu, Georgi Georgiev, Gabriela Marinova and Deniz Akinbosoye

July 2021

Abstract

K-Nearest Neighbors or shortly KNN is one of the least complex and widely used classification methods in Data Analysis. In this project, KNN is used to recognize and distinguish individual human faces in images. The classification algorithm uses the components of an image, which are obtained by performing Principal Component Analysis (PCA) on the dataset. The program receives a large folder of pictures divided into sub-folders, each subfolder containing a collection of pictures of a certain person. This data set is then divided into two parts, the testing set and the training set, and the purpose of the program is to determine in which sub-folder of the training set the images from the testing set belong. In the end, the accuracy of the program is calculated based on the number of correctly assigned pictures.