<![CDATA[This study explores the development of a face recognition model using the K-Nearest Neighbors (KNN) algorithm as a classification method in biometric systems. The dataset consists of 1,000 grayscale facial images sized 128x128 pixels, collected from 10 individuals with 100 images each. Feature extraction was conducted using the Histogram of Oriented Gradients (HOG) technique to capture distinctive facial characteristics. The experimental results show that the optimal k value is 4, producing a validation accuracy of 75.37%. Further testing achieved an accuracy of 81.37% with an average F1-score of 0.81, demonstrating reliable recognition performance. Live recognition tests confirmed that the system can still identify faces under real-world conditions, such as varied orientations and partial occlusions. These results indicate that KNN is an effective and efficient algorithm for small to medium-scale face recognition tasks, offering fast training time and practical applicability for biometric identification systems.]]>
Copyrights © 2026
