Implementation of Face Recognition Using Deep Metric Learning for Automatic Door Openers

The ever-increasing crime rate has fundamentally changed perceptions of home and office door security. In Indonesia, door security largely relies on traditional technologies such as keys and passcodes which although effective, are prone to issues like lost cards or forgotten passcodes. To address these challenges, this study investigates the adoption of facial recognition technology and IoT (Internet of Things). This technology leverages the Siamese neural network methodology, which consists of multiple identical sub-networks with uniform settings and weights, ensuring robust parameter adaptation across all components. This collective operational strategy enables the Siamese network to learn efficiently even with limited input data. The facial recognition results in the database will be used to open doors with the IoT system. This study was conducted using 500 facial images of students from the Surabaya State Electronics Polytechnic and achieved an impressive 88% accuracy rate in testing. By exploring facial recognition technology, this study aims to improve door security by overcoming the limitations of conventional methods, offering a promising alternative for a more secure and efficient access control system.