Developing a machine learning based face recognition attendance system for employees.

Author

Nhari, Natasha

Title

Developing a machine learning based face recognition attendance system for employees.

Abstract

This study presents the expansion and evaluation of a machine learning-based face recognition attendance system designed to update and streamline employee attendance tracking in organizations. Traditional attendance means such as manual registers and RFID cards are increasingly prone to issues such as inaccuracy, buddy punching, and administrative inefficiencies. To discourse these limitations, this study proposes a contactless, real-time facial recognition solution that ensures secure, accurate, and automated attendance recording. The system used advanced computer vision systems and machine learning algorithms, integrating facial detection and recognition modules using tools such as OpenCV and deep learning frameworks. Key image processing methods like Contrast Limited Adaptive Histogram Equalization (CLAHE), median filtering, and Principal Component Analysis (PCA) are used to enhance illumination, reduce noise, and extract relevant facial features. Euclidean distance is used as a metric to match detected faces with those stored in a secure database. Once a match is confirmed, attendance is automatically logged along with timestamp data. A prototype was developed and tested within a workplace setting. The system confirmed a promising performance in recognizing faces under variable conditions, with evaluation metrics including accuracy, precision, and recall used to assess its reliability. Feedback collected from employee surveys indicated a high level of user satisfaction, ease of use, and a general preference for the new system over traditional attendance methods. However, some challenges such as inconsistent lighting, limited hardware resources, and concerns over data privacy were noted. This research confirms that facial recognition technology, when properly implemented, offers a scalable, user-friendly, and efficient alternative to conventional attendance systems. It reduces manual workload, prevents time fraud, and supports a digital transformation in employee management practices. The system is particularly relevant in a post-pandemic era, where contactless technologies are increasingly vital. Future improvements may focus on enhancing recognition speed, strengthening data protection, and incorporating mobile and cloud-based functionality to extend the system’s reach and flexibility.

Date

June 2025

Publisher

BUSE

Keywords

Machine Learning

Face Recognition

Attendance System

Supervisor

Mr. Mhlanga