Rancang Bangun Smart Lock Door Menggunakan Face Recognition dengan Metode YOLO dan PIN sebagai MFA Berbasis Monitoring Telegram

Sabiilah, Rasyidien Mukti Barik Lana Nur (2025) Rancang Bangun Smart Lock Door Menggunakan Face Recognition dengan Metode YOLO dan PIN sebagai MFA Berbasis Monitoring Telegram. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 2038211050-Undergraduate_Thesis.pdf] Text
2038211050-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (8MB) | Request a copy

Abstract

Meningkatnya kasus kejahatan properti di Indonesia, khususnya di wilayah Jawa Timur, menunjukkan urgensi pengembangan sistem keamanan yang lebih cerdas. Penelitian ini bertujuan untuk merancang dan mengimplementasikan sistem kunci pintu cerdas (smart lock door) dengan autentikasi ganda yang mengintegrasikan pengenalan wajah (face recognition) menggunakan metode YOLOv5 (You Only Look Once) dan masukan PIN melalui papan tombol (keypad), serta merancang sistem notifikasi berbasis Telegram. Sistem ini dikembangkan menggunakan Raspberry Pi 5 sebagai unit pemrosesan utama, modul kamera Raspberry Pi v1.3 untuk pengambilan citra wajah, keypad 4x4 sebagai alat autentikasi PIN, serta modul relay 5V satu chanel low trigger dan kunci solenoid 12V untuk pengendalian akses pintu, dengan catu daya untuk solenoid berupa baterai Li-On SH1933. Metode pengujian dilakukan pada tiga variasi jarak (30 cm, 45 cm, dan 60 cm) dalam kondisi pencahayaan terang, dengan total enam subjek pengujian yang terdiri dari tiga subjek dikenal (termasuk dalam data pelatihan) dan tiga subjek tidak dikenal (tidak termasuk dalam data pelatihan). Subjek tidak dikenal diidentifikasi melalui nilai confidence di bawah 0,85 dari model deteksi wajah. Hasil pengujian menunjukkan bahwa sistem mampu mencapai akurasi pengenalan wajah sebesar 96,67%, dengan nilai precision sebesar 0,93, recall sebesar 1,00, dan F1 Score sebesar 0,96. Waktu respons notifikasi Telegram tercatat selama 2,73 detik pada wajah yang tidak dikenali, 22,27 detik pada notifikasi kesalahan autentikasi sebanyak tiga kali, dan 19,49 detik pada kondisi kunci solenoid terbuka setelah seluruh proses autentikasi berhasil dilakukan. Dengan performa tersebut, sistem ini terbukti mampu memberikan keamanan berlapis melalui metode autentikasi ganda dan pemantauan jarak jauh secara waktu nyata, serta berpotensi diterapkan dalam lingkungan rumah pintar (smart home) atau sistem akses terbatas yang memerlukan keamanan berlapis.
===================================================================================================================================
The increasing rate of property crimes in Indonesia, particularly in East Java, highlights the urgency of developing more smart security systems. This project aims to design and implement a smart lock door system with multi-factor authentication that integrates face recognition using the YOLOv5 (You Only Look Once) method and PIN input via a keypad, along with a Telegram-based notification system. The system is developed using a Raspberry Pi 5 as the main processing unit, a Raspberry Pi Camera Module v1.3 for face image capture, a 4x4 keypad for PIN authentication, and a 5V single-channel low-trigger relay module and a 12V solenoid lock for door access control. The solenoid is powered by a Li-On SH1933 battery. The testing method involved three distance variations (30 cm, 45 cm, and 60 cm) under bright lighting conditions, with a total of six test subjects: three known subjects (included in the training dataset) and three unknown subjects (not included in the dataset). Unknown subjects were identified based on a confidence Score below 0.85 from the face detection model. The test results showed that the system achieved a face recognition accuracy of 96,67%, with a precision of 0,93, a recall of 1,00, and an F1 Score of 0,96. Telegram notification response times were recorded as 2,73 seconds for unrecognized faces, 22,27 seconds for three consecutive incorrect PIN attempts, and 19,49 seconds for successful authentication and solenoid lock activation. With these performance results, the system proves capable of providing layered security through multi-factor authentication and real time remote monitoring, making it suitable for smart home environments or restricted access systems requiring enhanced security.

Item Type: Thesis (Other)
Uncontrolled Keywords: Raspberry Pi 5, YOLOv5, Autentikasi Ganda, Pengenalan Wajah, Notifikasi Telegram , Multi-Factor Authentication, Face Recognition, Telegram Notification
Subjects: T Technology > TJ Mechanical engineering and machinery > TJ211 Robotics.
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5103.2 Wireless communication systems. Two way wireless communication
Divisions: Faculty of Vocational > Mechanical Industrial Engineering (D4)
Depositing User: Rasyidien Mukti Barik Lana Nur Sabiilah
Date Deposited: 30 Jul 2025 01:43
Last Modified: 31 Jul 2025 03:37
URI: http://repository.its.ac.id/id/eprint/123395

Actions (login required)

View Item View Item