Sistem Keamanan Menggunakan Pengenalan Wajah Untuk Fasilitas Umum Berbasis Gender

Waisnawa, I Gede Kadek Restu Kartana (2024) Sistem Keamanan Menggunakan Pengenalan Wajah Untuk Fasilitas Umum Berbasis Gender. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022201245-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (6MB) | Request a copy

Abstract

Aspek keamanan pada fasilitas umum menjadi hal yang krusial untuk ditingkatkan, mengingat semakin maraknya insiden kejahatan seperti kasus penjambretan, kekerasan fisik pelecehan seksual, dsb. Lebih-lebih dewasa ini banyak sekali pemberitaan tentang kasus-kasus pelecehan seksual pada wanita justru terjadi di fasilitas umum tertentu yang berbasis gender. Tantangan dalam menjaga keamanan masyarakat umum membutuhkan solusi yang inovatif dengan pemanfaatan teknologi terbaru seperti Artificial Intelligence (AI) yang bisa diimplementasikan pada sistem keamanan yang berbasis teknologi. Dalam tugas akhir ini telah dibuat desain dan realisasi sistem keamanan berbasis gender bernama Gender Oversight System (GENOS) yang mengintegerasikan antara AI, kamera CCTV, Mikrokontroller, dan sistem respon. Sistem respon dipilih berupa suara alarm apabila terdeteksi laki-laki, dan menyalakan indikator lampu apabila terdeteksi perempuan. Sistem GENOS dalam tugas akhir ini pada dasarnya terdiri dari 3 bagian yaitu Akuisisi Citra, Sistem Deteksi Gender (SDG), dan Sistem Respon. Adapun untuk merealisasikan prototipe sistem GENOS tersebut, digunakan pendekatan deep learning metode Convolutional Neural Network (CNN) dan transfer learning, pada bagian sistem pengenalan wajah. Ada beberapa library yang dilibatkan dalam program GENOS yaitu Tensorflow, OpenCV, Numpy, dan Pandas. Dalam sisi hardware, digunakan perangkat CCTV (Ezfiz C6N) untuk menangkap wajah secara real time, laptop untuk mengolah data wajah, dan mikrokontroller (Arduino UNO) untuk mengatur kerja respon sistem. Respon sistem sendiri terdiri dari komponen Buzzer Aktif (5V) dan LED (5V). Pengujian terhadap program Sistem Deteksi Gender (SDG) yang dibuat dalam tugas akhir ini, dilakukan menggunakan Dataset UTK-Face dan Metode-1 (CNN) dan Metode-2 (CNN + Transfer Learning). Hasil terbaik diperoleh pada Metode-2 dengan nilai performansi 96.13% dan tingkat kesalahan 3.83%. Training-model dari hasil pengujian tersebut selanjutnya diimplementasikan pada sistem GENOS. Setelah dilakukan pengujian menggunakan dua skenario dataset yang berbeda (Citra Pas Foto & Citra CCTV), maka secara keseluruhan sistem GENOS memiliki kinerja sebesar 92.92% dan tingkat kesalahan 7%. Kinerja ini dianggap telah memenuhi target yang direncanakan mengacu pada standar desain dengan nilai akurasi > 85% dan tingkat loss < 25%.
=================================================================================================================================
The security aspect of public facilities has become crucial to improve, given the increasing incidents of crimes such as robbery, physical violence, sexual harassment, etc. Recently, there have been numerous reports of sexual harassment cases against women occurring in certain gender-based public facilities. The challenge of maintaining public safety requires innovative solutions utilizing the latest technology, such as Artificial Intelligence (AI), which can be implemented in technology-based security systems. In this thesis, a gender-based security system called Gender Oversight System (GENOS) was designed and realized, integrating AI, CCTV cameras, microcontrollers, and a response system. The response system was designed to trigger an alarm sound when a male is detected and to activate a light indicator when a female is detected. The GENOS system in this thesis essentially consists of three parts: Image Acquisition, Gender Detection System (SDG), and Response System. To realize the GENOS prototype, a deep learning approach using Convolutional Neural Network (CNN) and transfer learning methods was employed for the facial recognition system. Several libraries were involved in the GENOS program, including TensorFlow, OpenCV, Numpy, and Pandas. On the hardware side, a CCTV device (Ezviz C6N) was used to capture real-time facial images, a laptop for processing facial data, and a microcontroller (Arduino UNO) to control the response system. The response system itself consists of an Active Buzzer (5V) and an LED (5V). Testing of the Gender Detection System (SDG) program developed in this thesis was conducted using the UTK-Face Dataset with Method-1 (CNN) and Method-2 (CNN + Transfer Learning). The best results were obtained with Method-2, achieving a performance score of 96.13% and an error rate of 3.83%. The trained model from this testing was then implemented in the GENOS system. After testing with two different dataset scenarios (Passport Photo Images & CCTV Images), the overall performance of the GENOS system was 92.92% with an error rate of 7%. This performance is considered to have met the planned target, referring to the design standard with an accuracy of > 85% and a loss rate of < 25%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Convolutional Neural Network (CNN), Deteksi Gender, Kamera Keamanan, RTSP, Deep Learning, Gender Detection, Image Processing, Security Camera.
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques T Technology > TA Engineering (General). Civil engineering (General) > TA1650 Face recognition. Optical pattern recognition. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	I GEDE KADEK RESTU KARTANA WAISNAWA
Date Deposited:	29 Jul 2024 08:31
Last Modified:	29 Jul 2024 08:31
URI:	http://repository.its.ac.id/id/eprint/109833

Actions (login required)

View Item