DESIGN AND DEVELOPMENT OF A FACE RECOGNITION PROTOTYPE BASED ON YOLO11 USING RASPBERRY PI

Abstract

This research aims to design a face recognition prototype based on YOLO11 using a Raspberry Pi. The prototype design employs a Raspberry Pi 4B, with input from the Raspberry Pi Camera Module 2 and output in the form of audio that provides identification results based on face recognition. In this study, a face dataset consisting of 250 photos with 10 classes or labels was used, meaning that the prototype can recognize faces from 10 individuals. The dataset was divided into 80% (200) face images for training, 8% (20) for testing, and 12% (30) for validation. Based on the testing results of 10 moving videos, the prototype achieved an accuracy of 95%, precision of 100%, recall of 89.6%, and an F1-Score of 94%. Nevertheless, the identification performance is sensitive to backlight conditions, motion blur, and extreme head poses, which can reduce detection accuracy. The Task Success Rate testing for measuring speaker performance reached 100%, indicatin