Implementation of a Facial Recognition Attendance System using Raspberry Pi

Abstract

Manual attendance systems, such as manual sign-in sheets in academic institutions, are often error-prone, time-consuming, and can be easily compromised. Fingerprint biometric systems can pose hygiene risks due to shared contact surfaces. This motivated the development of an automated, contactless facial recognition-based attendance system to enhance accuracy, efficiency, and hygiene in educational settings. This project aimed to design and implement a system that uses facial recognition technology to accurately log student attendance in real time, while leveraging the use of the Raspberry Pi. The system integrates a Raspberry Pi Camera Module for capturing facial images, a 16x2 I2C LCD for displaying recognized names and matric numbers, and a buzzer for auditory confirmation. Using Python, the system was built with key libraries such as face recognition, dlib, and OpenCV. A dataset comprising 60 students, each with about 15 facial images, was used for training. Facial features were extracted using 128-dimensional embedding from dlib and classified with a Support Vector Machine (SVM). The system was tested under different lighting and angles, which resulted in producing an accuracy range of 83–96%, an average recognition time of 6 seconds, a false acceptance rate (FAR) of 15%, and a false rejection rate (FRR) of 10%. This paper is subdivided into 6 sections. Section 1 is the introduction, Section 2 is the literature review, Section 3 is Methodology, Section 4 results and discussion, Section 5 recommendations, Section 6 conclusion.