Face Recognition System with Deep Learning for Automated Workspace Access

Abstract

Conventional access control systems, such as access cards and PIN codes, have long been utilized to secure high-security workspaces. However, these traditional methods are increasingly vulnerable to forgery, theft, and unauthorized sharing, posing significant security risks. The limitations of manual systems such as stolen cards and password leaks highlight the need for more secure and efficient alternatives. This study aims to enhance workplace security by implementing a face recognition system, offering a more secure and efficient method of access control that is less susceptible to fraud.Face recognition technology has evolved significantly with advancements in deep learning, particularly Convolutional Neural Networks (CNN) and FaceNet embedding. These techniques allow for more accurate and reliable face recognition, even under challenging conditions like varying lighting, different head poses, and occlusions. The proposed system uses CNN architecture to extract facial features and FaceNet embedding to create facial feature vectors, which are highly discriminative and robust. The dataset used for training includes facial images captured under diverse environmental conditions, with preprocessing techniques applied to ensure effective recognition.The face recognition system achieved a 95% identification success rate, demonstrating its robustness and reliability. In comparison to traditional methods, the face recognition system offers enhanced security, as it is resistant to spoofing attacks and does not require physical tokens, reducing the risk of unauthorized access. Additionally, face recognition is more user-friendly and hygienic compared to fingerprint or iris scanning systems. The proposed system is also cost-effective and easy to implement, particularly in large-scale environments.Future improvements could include real-time monitoring, integration with additional security measures, and exploration of adaptive learning techniques to further enhance the system’s performance and robustness in dynamic environments.