Implementation of a Face Recognition-Based Door Access Control System for Corporate Offices in Lagos

Physical access control in corporate offices in Lagos, South West Nigeria, relies predominantly on legacy key-card or manual security systems that are vulnerable to card-cloning and social engineering. This project implemented a face recognition-based door access control system using a Raspberry Pi 4 and a camera module. The system captured facial images at the entry point, applied the OpenCV Haar Cascade detector for face localisation, and used the LBPH (Local Binary Pattern Histogram) algorithm for identity recognition against an enrolled database of authorised personnel. On successful recognition, a relay activated an electric door strike solenoid. Failed recognition triggered an alert buzzer and logged the attempt with a timestamped image to local storage. The system was enrolled with 20 authorised users, each with 30 training images captured under varied lighting conditions. Recognition accuracy tested across 500 identification attempts was 94.6 percent. False acceptance rate was 2.1 percent and false rejection rate was 3.3 percent. System response time from face detection to door actuation averaged 1.8 seconds. Image logging successfully captured and stored entry events with 100 percent reliability. Power consumption was 4.2W in standby and 6.1W during active recognition. The study acknowledges accuracy degradation under very low ambient light conditions and recommends integrating an infrared illuminator for consistent night-time performance, alongside encrypted cloud backup of access logs for compliance with corporate security audit requirements.

Keywords: face recognition, access control, OpenCV, Raspberry Pi, Lagos