Main Article Content
Abstract
This study presents the development and performance evaluation of an airport security gate prototype based on facial recognition technology to enhance verification efficiency and minimize human error in passenger identity checks. The proposed system utilizes Raspberry Pi 4 as the main embedded controller, integrated with a USB camera for image acquisition, MTCNN (Multi-Task Cascaded Convolutional Networks) for robust face detection, FaceNet pretrained model for deep feature extraction, and Support Vector Machine (SVM) for identity classification. Output components include an I2C LCD module for real-time status display and a servo motor for automated gate access control. Experimental testing was conducted on five enrolled individuals with a dataset of 700 facial images (224×224 pixels) captured under varying lighting conditions. Results demonstrated recognition accuracy of 96–98% under optimal lighting conditions (108 lux) with an average response time of 5–7 seconds. However, system performance significantly degraded under low illumination (37 lux) and non-frontal face positions, indicating environmental sensitivity. The closed-loop architecture with common ground implementation ensured stable stand-alone operation. This prototype offers a low-cost, automated alternative to conventional manual verification systems, though further development is required to enhance robustness in diverse environmental conditions and enable real-time enrollment capabilities.