Implementasi Sistem Pengenalan Wajah Pada Keamanan Akses Masuk Warehouse and Bagging Command Center Menggunakan Metode Convolutional Neural Network (CNN)

Fajriati, Arneta (2025) Implementasi Sistem Pengenalan Wajah Pada Keamanan Akses Masuk Warehouse and Bagging Command Center Menggunakan Metode Convolutional Neural Network (CNN). Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 2040211088-Undergraduate _Thesis.pdf] Text
2040211088-Undergraduate _Thesis.pdf
Restricted to Repository staff only

Download (7MB) | Request a copy

Abstract

Keamanan akses pintu sangat penting dalam melindungi aset dan data perusahaan, terutama di area dengan aktivitas operasional dan penyimpanan barang bernilai tinggi. Berdasarkan data operasional, rata-rata sekitar 10-40 orang keluar masuk kantor Command Center setiap harinya, termasuk karyawan, teknisi, dan pihak eksternal dengan izin akses khusus. Tingginya volume aktivitas ini membuat pengawasan dan kontrol akses secara manual menjadi lebih sulit dan rentan terhadap kesalahan manusia. Penelitian ini bertujuan untuk mengimplementasikan sistem pengenalan wajah berbasis Convolutional Neural Network (CNN) khususnya FaceNet sebagai solusi peningkatan keamanan akses pintu masuk pada kantor Warehouse and Bagging Command Center PT Petrokimia Gresik. Sistem dirancang agar dapat mengenali wajah karyawan yang terdaftar dan menolak akses bagi wajah yang tidak dikenali. Metode yang digunakan dalam penelitian ini melibatkan model FaceNet yang dilatih menggunakan fungsi Triplet Loss, dengan pendekatan semi-hard triplet mining untuk menghasilkan representasi embedding wajah dalam bentuk vektor berdimensi 128. Proses klasifikasi dilakukan dengan menghitung jarak Euclidean antara vektor wajah input dan vektor dalam database. Sistem ini dirancang untuk meningkatkan efisiensi dan akurasi kontrol akses pintu dengan memberikan otorisasi otomatis berdasarkan identifikasi wajah, sehingga hanya orang yang terdaftar yang dapat mengakses area sensitif tersebut. Hasil pengujian sistem menunjukkan bahwa sistem pengenalan wajah berbasis CNN khususnya FaceNet dapat mengenali wajah dengan tingkat akurasi mencapai 91% dari total 3750 data dengan 15 label wajah berbeda. Pengujian dilakukan dengan berbagai kondisi, termasuk pencahayaan yang berbeda (terang dan gelap), ekspresi wajah (normal dan senyum), serta posisi wajah (miring kiri dan kanan). Sistem ini dapat bekerja dengan baik dalam kondisi tersebut, dengan akurasi pengenalan tertinggi pada threshold 0,5. Selain itu, pengujian terhadap perangkat keras, seperti Raspberry Pi, solenoid door lock, dan relay, menunjukkan bahwa sistem bekerja secara otomatis dan efisien dalam mengontrol akses pintu, mengurangi ketergantungan pada pengawasan manual, dan meminimalkan kesalahan manusia.
==============================================================================================================================
Door access security is crucial in protecting company assets and data, especially in areas with operational activities and the storage of high-value items. According to operational data, an average of 10-40 people enter and exit the Command Center office daily, including employees, technicians, and external parties with special access permissions. The high volume of activity makes manual monitoring and access control more difficult and prone to human error. This research aims to implement a face recognition system based on Convolutional Neural Network (CNN), specifically FaceNet, as a solution to improve door access security at the Warehouse and Bagging Command Center of PT Petrokimia Gresik. The system is designed to recognize the faces of registered employees and deny access to unrecognized faces. The methodology used in this research involves the FaceNet model, which is trained using the Triplet Loss function, with a semi-hard triplet mining approach to generate a 128-dimensional face embedding vector. The classification process is carried out by calculating the Euclidean distance between the input face vector and the vectors in the database. This system is designed to enhance the efficiency and accuracy of door access control by providing automatic authorization based on facial identification, ensuring that only registered individuals can access sensitive areas. The system's test results show that the CNN-based face recognition system, particularly FaceNet, can recognize faces with an accuracy of up to 91%. The tests were conducted under various conditions, including different lighting (bright and dark), facial expressions (normal and smiling), and face positions (tilted left and right). The system performs well under these conditions, with the highest recognition accuracy at a threshold of 0.5. Additionally, hardware testing on components such as Raspberry Pi, solenoid door lock, and relay showed that the system works automatically and efficiently in controlling door access, reducing reliance on manual monitoring, and minimizing human errors.

Item Type: Thesis (Other)
Uncontrolled Keywords: Keamanan Akses Pintu, Convolutional Neural Network (CNN), FaceNet,Triplet Loss. Door Access Security, Convolutional Neural Network (CNN), FaceNet, Triplet Loss
Subjects: T Technology > T Technology (General)
T Technology > T Technology (General) > T57.5 Data Processing
Divisions: Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User: Arneta Fajriati
Date Deposited: 07 Aug 2025 09:05
Last Modified: 07 Aug 2025 09:05
URI: http://repository.its.ac.id/id/eprint/127967

Actions (login required)

View Item View Item