Analisis Kinerja Algoritma K-Nearest Neighbors untuk Identifikasi dan Pengenalan Wajah

Abstract

Development of science and technology modern face detection and recognition applied in more sectors like verification identity by every face move application from various orientation angles. K-Nearest Neighbors algorithm is a classification object method according to data learning that has a near distance from the tested object. Using 50 captured data training every object from the tested object use data images and a real-time webcam. Tested using pictures to get good results accuracy is 0.94. While tested with by real-time webcam gets different results from various tested objects. The first object gets accuracy value is 0.83, second object gest an accuracy value is 0.59. The analysis tested performance algorithm K-Nearest Neighbors results for face identification and recognition are proven to be able to produce good accuracy values for flexible data.