The Office Room Security System Using Face Recognition Based on Viola-Jones Algorithm and RBFN Sistem Keamanan Ruang Kantor Menggunakan Fitur Pengenalan Wajah Berbasis Algoritma Viola-Jones dan RBFN

The university as an educational institution can apply technology in the campus environment. Currently, the security system for office space that is integrated with digital data has been somewhat limited. The main problem is that office space security items are not guaranteed as there might be outsiders who can enter the office. Therefore, this study aims to develop a system using biometric (face) recognition based on Viola-Jones and Radial Basis Function Network (RBFN) algorithm to ensure office room security. Based on the results, the system developed shows that object detection can work well with an object detection rate of 80%. This system has a pretty good accuracy because the object matching success is 73% of the object detected. The final result obtained from this study is a prototype development for office security using face recognition features that are useful to improve safety and comfort for occupants of office space (due to the availability of access rights) so that not everyone can enter the office. Keyword—Office Security, Face Recognition, Prototyping, Database Abstrak—Universitas sebagai salah satu institusi pendidikan dapat menerapkan teknologi untuk meningkatkan keamanan kantor dosen dan staf. Pada saat ini, sistem keamanan kantor dosen dan staf yang terintegrasi dengan data dosen ataupun staf akademik secara digital masih terbatas, sehingga ditemukan kendala utama yaitu berupa keamanan barang-barang penting dan bersifat rahasia tidak terjamin karena dengan bebasnya pihak luar masuk ke dalam ruang kantor. Oleh karena itu, penelitian ini bertujuan untuk mengembangkan sistem keamanan ruang kantor menggunakan fitur pengenalan wajah berbasiskan algoritma Viola-Jones dan Radial Basis Function Network (RBFN). Berdasarkan hasil pengujian, sistem yang dibuat menunjukkan pendeteksian objek dapat berjalan dengan cukup baik dengan tingkat pendeteksian objek individu 80%, dan sistem ini memiliki akurasi yang cukup baik karena keberhasilan mencocokan objek individu yang terdeteksi mencapai 73% dengan data objek individu yang tersimpan dalam database. Hasil akhir yang diperoleh dari penelitian adalah prototype sistem keamanan ruang kantor menggunakan fitur pengenalan wajah yang bermanfaat untuk meningkatkan keamanan dan kenyaman bagi penghuni ruang kantor, karena adanya hak akses ke dalam ruang kantor, sehingga tidak semua orang dapat memasuki ruang kantor tersebut. Kata Kunci—Sistem Keamanan Ruangan, Pengenalan Wajah, Prototyping, Database INTENSIF, Vol.5 No.1 February 2021 ISSN: 2580-409X (Print) / 2549-6824 (Online) DOI: https://doi.org/10.29407/intensif.v5i1.14435 2 INTENSIF: Jurnal Ilmiah Penelitian dan Penerapan Teknologi Sistem Informasi


I. INTRODUCTION
In today's modern era, the security aspect is essential because a high-security level can increase individuals' peace and comfort [1]. People can do various ways to improve security, ranging from the direct or physical human involvement in the form of guarding by security officers to using technology support, for example, information security systems [2][3], home security systems [4], digital rights management [5], and data distribution security systems [6].
Also, the use of technology in the security system applied can be the use of biometrics such as fingerprints, CCTV installation, and various other things [7].
As an educational institution, the university can significantly apply technology in the campus environment to increase the office space's security (workspace) of lecturers and staff. Based on observations in the campus environment, few office security systems are currently integrated with digital data from lecturers or academic staff. Therefore the main obstacle is possible, namely the security of essential and confidential items that are not guaranteed security because outsiders enter office room freely.
At this time, the application of room security systems can be carried out with technological support in the form of the Internet of Things (IoT) using Arduino hardware, integrated with a database to maximize room security systems [8]. The use of Arduino is increasing with the need for a microcontroller device that is affordable and supported by the development of work automation using a device that can be controlled via a mobile application [9]. Several previous studies regarding security systems have been carried out by M. K. Syabibi and A. Subari, [10], namely by using a webcam to create a web-based monitoring system for home security. Previous research aims to monitor activities that occur at home via the web. In contrast, this research seeks to create a security system for office doors, especially to control access of people in and out of the room.
Another research was carried out by M. Arihutomo [11] to make a guard robot for guarding.
The difference with this research is that previous research made robots a security system, whereas this study produced a security system in-room access rights. Other facial recognition research was also carried out by Yuliana Y. and Nurhaida I. [12]. The analysis uses a webcam to detect and perform facial recognition and tests to improve a person's face's detection and recognition accuracy. The research that is currently being conducted aims to implement an office security system using facial recognition features. Furthermore, several previous studies related to this research topic are summarized in  [13] developed a facial biometrics system to recognize faces with an accuracy rate of up to 97.5%. This study used the Gabor KPCA method and Mahalanobis Distance.
2 N. Saubari, R. Isnanto, and K. Adi [14] developed a face detection system using the Haar-Like Feature method and Artificial Neural Network Backpropagation. The facial recognition accuracy of the developed system reaches 80.8%. 3 I, K. S. Widiakumara, I K. G. D. Putra, and K. S. Wibawa [15] developed an Android application to identify faces. The method used is the Eigenface method, which is stored in the MySQL database system. The accuracy (success) of this facial recognition Android application is 68%. 4 N. V. de Lima, L. Novamizanti, and E. Susatio [16] developed a 3-dimensional facial recognition identification system based on template matching, the Iterative Closest Point (ICP) method, and the Support Vector Machine (SVM) classification.

A. Research Stages
This study uses the prototyping method in developing office space security systems. The stages in this research are shown in Figure 1 below. This research consisted of 3 phases, namely, requirement gathering, the second phase building a prototyping system, and system deployment for the final step. In this prototyping system, the user will try the system to get feedback and become an evaluation to produce prototype improvements. This process is repeated until analysts, users, and related parties agree on the resulting prototype [19].

B. Viola-Jones Algorithm
Face detection is an early stage in facial recognition systems. One of the algorithms that can be used to detect objects is the Viola-Jones algorithm. The Viola-Jones algorithm works by using a simple Haar-like feature that quickly evaluates a new image representation [20]. Viola-Jones processes feature sets with integral imagery and boost algorithms to reduce time complexity and perform a classifier's cascade classifier. The detection is carried out using the Viola-Jones algorithm, as shown in Figure 2. hidden layer (hidden layer), and output layer (output layer) [11]. The input layer receives an input vector A, which is then taken to the hidden layer, which will process the input data nonlinearly with the activation function. The output from the hidden layer is then processed linearly INTENSIF, Vol. at the output layer [21]. Radial Basis Function Network architectural modeling is seen in (Figure 3).  The block diagram of the office security system built is presented in Figure 4.In the diagram shown, two processes are the training process and the facial recognition process.

A. Research Prototyping Results
After conducting this research, the results obtained are in the form of a prototype of an office security system using facial recognition features. The structure built is a pilot project of the smart campus master plan that will be developed. The resulting prototype uses an Arduino NodeMCU ESP8266 microcontroller, wireless router, electronic door lock access, CPU server, and IP camera. How the office security system works using the facial recognition feature can be seen in Figure 5. The workings of the resulting prototype, namely: 1. When someone approaches the office door with a distance of about 1-2 meters, the IP camera will detect individual objects in front of them.
2. Then the captured individual objects will be sent via the wireless network to the CPU Server for matching individual faces captured by the IP camera with personal object data that has been stored in the database with the Radial Basis Function Network. Next, we will discuss the documentation for implementing a prototype office security system using the facial recognition feature. The office space security system created is then implemented for testing the system being built. Figure 6 shows the office space security system prototype in the form of an Arduino NodeMCU ESP8266 device and an IP Camera that has been completed and is connected to a wifi network and a CPU Server.   (Figure 8), when someone approaches the office door, the IP camera will detect individual objects in front of them. Furthermore, in Figure 9, the room security system will carry out the facial recognition process before the individual opens the door.    In Figure 13, the person in front of the door opens the door, enters the faculty and staff office room, and then closes the door. In Figure 14, the closed door will be automatically locked so that the office space's security is well maintained. In this office space security system research, there are two main processes: detecting objects by the camera at a distance of 1-2 meters and matching individual items captured by the camera with personal object data stored in the database. Tests are carried out on the Arduino-based office security system using the facial recognition feature to ensure that it runs well and is stable. The office space security system test was conducted 30 times to determine its ability to detect objects. The results of testing the office security system to detect individual objects are seen in Table 2: After the first stage of testing is carried out, the second stage of testing is carried out. Namely, the office space security system matches the individual objects detected with the personal object data stored in the database. This office space security system test was carried out 30 times to determine the system's ability to match individual objects' faces that have been detected. You can see the results of testing the office security system to fit the detected particular objects in Table 3. It can overcome the current obstacle by increasing the lighting in the room for maximum light intensity. The camera can get individual objects properly and pay attention to the object's angle by adding several cameras to cover a wider area and angle. Besides, to produce better results, researchers can pay attention to the image extraction process. The improved image extraction process can help the recognition process to be more accurate [13]. It is estimated that the next researchers will solve this problem by developing a more modern face detection algorithm with a higher accuracy level.

IV. CONCLUSION
This research concludes that the office space security system prototype using facial recognition features works well (80% successful detection of individual objects). As for the face matching process, the office space security system developed has good accuracy because matching objects' success reaches 73%. Suggestions for future research are that in implementing an office security system that uses facial recognition features, it is necessary to apply a combination of methods and other face recognition algorithms to increase accuracy and fast processing times. This research concludes that facial recognition technology is currently still under development by experts to obtain a high level of accuracy. People can maximally realize the result of smart living, smart city, smart home, and intelligent campus.