Penerapan Face Recognition dengan Model Deep Neural Network pada Sistem Smart Door Lock

The advancements in artificial intelligence and IoT have created significant opportunities to enhance home security systems. This research proposes a Smart Door Lock system based on Face Recognition, utilizing a Deep Neural Network (DNN) integrated with an ESP32-CAM and a Solenoid Door Lock. The system employs MTCNN for face detection and compares facial embeddings using Euclidean Distance, along with adjustments to the similarity threshold. Testing results demonstrate that the system achieves an average accuracy of 84%, with evaluation metrics of 95% Precision, 85% Recall, and an F1-Score of 89.72%. Adjusting the similarity threshold from 0.60 to 0.28 significantly reduces the false positive rate—from 7 to 1 under certain conditions—although overall tests indicate a false positive rate of approximately 5%. Additionally, the system logs access events, enabling homeowners to monitor door access in real time.