This paper presents the design and implementation of an embedded facial recognition access control system developed on a Raspberry Pi 4, aimed at enhancing security in institutional or restricted-access environments. The proposed solution integrates real-time facial detection using the Haar Cascade classifier and facial recognition through the Local Binary Pattern Histograms (LBPH) algorithm, both widely recognized for their balance between computational efficiency and accuracy in constrained hardware. The system is operated through a graphical user interface built with Tkinter, enabling intuitive user interaction for system configuration and operation. Key functionalities include user registration, training of the facial recognition model, generation and export of access reports, and physical activation of an electromagnetic lock to control door entry. During experimental validation under controlled lighting and positioning conditions, the system achieved a facial recognition accuracy exceeding \(87\%\) , with an average identification response time of 1.4 s and a total access time of approximately 6.5 s. These results underscore the feasibility of deploying low-cost, embedded biometric systems for security-critical applications, particularly in environments with limited resources.