EVALUASI METODE FACE RECOGNITION MENGGUNAKAN METODE FISHER’S LINEAR DISCRIMINANT (FLD) UNTUK APLIKASI LOCK SYSTEM PADA SMART HOME

Ibrahim, Kaisar Malik (2023) EVALUASI METODE FACE RECOGNITION MENGGUNAKAN METODE FISHER’S LINEAR DISCRIMINANT (FLD) UNTUK APLIKASI LOCK SYSTEM PADA SMART HOME. S2 thesis, Universitas Mercu Buana - Menteng.

	Text (COVER) 55420110016-Kaisar Malik Ibrahim-01 Cover - Malik Ibrahim.pdf Download (463kB)
	Text (BAB I) 55420110016-Kaisar Malik Ibrahim-03 Bab 1 - Malik Ibrahim.pdf Restricted to Registered users only Download (135kB)
	Text (BAB II) 55420110016-Kaisar Malik Ibrahim-04 Bab 2 - Malik Ibrahim.pdf Restricted to Registered users only Download (199kB)
	Text (BAB III) 55420110016-Kaisar Malik Ibrahim-05 Bab 3 - Malik Ibrahim.pdf Restricted to Registered users only Download (174kB)
	Text (BAB IV) 55420110016-Kaisar Malik Ibrahim-06 Bab 4 - Malik Ibrahim.pdf Restricted to Registered users only Download (872kB)
	Text (BAB V) 55420110016-Kaisar malik Ibrahim-07 Bab 5 - Malik Ibrahim.pdf Restricted to Registered users only Download (41kB)
	Text (DAFAR PUSTAKA) 55420110016-Kaisar Malik Ibrahim-09 Daftar Pustaka - Malik Ibrahim.pdf Restricted to Registered users only Download (72kB)
	Text (LAMPIRAN) 55420110016-Kaisar Malik Ibrahim-10 Lampiran - Malik Ibrahim.pdf Restricted to Registered users only Download (1MB)

Abstract

Peningkatan jumlah penduduk diiringi dengan peningkatan persentase kejahatan seperti tindak pencurian pada rumah. Penelitian ini bertujuan untuk menguji efektivitas metode Fisher's Linear Discriminant (FLD) dalam rancangan sistem pengenalan wajah pada lock system di smart home serta mengukur tingkat keamanan yang dihasilkan. Metode analisis diskriminan dengan FLD digunakan untuk memproses data dan menghasilkan evaluasi akurat. Hasil penelitian menunjukkan bahwa tingkat keamanan sistem pengamanan kunci pada smart home sangat dipengaruhi oleh data set latih. Semakin besar jumlah data set latih yang digunakan, semakin tinggi keamanan sistem. Namun, uji coba juga mengungkapkan bahwa identifikasi wajah dengan FLD dipengaruhi oleh penggunaan aksesorisseperti kacamata hitam dan masker, menyebabkan penurunan akurasi. Hasil evaluasi menunjukkan bahwa FLD memiliki tingkat akurasi 100% dalam mendeteksi wajah tanpa aksesoris. Namun, pada kondisi penggunaan kacamata hitam, tingkat akurasi deteksi wajah menurun menjadi 92,86%. Penggunaan masker menyebabkan tingkat akurasi deteksi wajah yang rendah,hanya 57,14%. Masker menyembunyikan fitur wajah, menghambat proses deteksi,dan menjadi kendala dalam pengenalan wajah. FLD efektif digunakan dalam situasi umum tanpa aksesoris, tetapi penggunaan kacamata hitam dan masker mempengaruhi tingkat akurasi deteksi wajah. Dalam merancang sistem face recognition untuk pengamanan kunci pada smart home, diperlukan perhatian khusus terhadap kondisi penggunaan aksesoris untuk meningkatkan keamanan dan kinerja sistem. The increase in population is accompanied by an increase in the percentage of crimes such as home theft. This research aims to test the effectiveness of the Fisher's Linear Discriminant (FLD) method in designing facial recognition systems for lock systems in smart homes and measure the resulting level of security. The discriminant analysis method with FLD is used to process data and produce accurate evaluations. The research results show that the security level of the key security system in a smart home is greatly influenced by the training data set. The larger the number of training set data used, the higher the security of the system. However, trials also revealed that facial identification with FLD was affected by the use of accessories such as sunglasses and masks, causing a decrease in accuracy. The evaluation results show that FLD has a 100% accuracy rate in detecting faces without accessories. However, when using sunglasses, the level of face detection accuracy decreases to 92.86%. The use of masks causes a low level of face detection accuracy, only 57.14%. Masks hide facial features, hinder the detection process, and become obstacles in facial recognition. FLD is effective in general situations without accessories, but the use of sunglasses and masks affects the accuracy of face detection. In designing a facial recognition system to secure keys in a smart home, special attention is needed to the conditions of use of accessories to improve security and system performance.

Actions (login required)

View Item