Convolutional Neural Networks Untuk Pengenalan Wajah Secara Real-Time

Zufar, Muhammad (2016) Convolutional Neural Networks Untuk Pengenalan Wajah Secara Real-Time. Undergraduate thesis, Institut Technology Sepuluh Nopember.

[img] Text
1212100076-Undergraduate Thesis.pdf - Published Version

Download (2MB)

Abstract

Identifikasi identitas individu melalui pengenalan wajah secara otomatis merupakan suatu persoalan besar yang menarik dan banyak sekali berbagai macam pendekatan untuk menyelesaikan persoalan ini. Apalagi di dalam skenario kehidupan nyata yang tidak terkontrol, wajah akan terlihat dari berbagai sisi dan tidak selalu menghadap ke depan yang membuat permasalahan klasifikasi menjadi lebih sulit diselesaikan. Dalam Tugas Akhir ini digunakan salah satu metode deep neural networks yaitu Convolutional Neural Networks (CNN) sebagai pengenalan wajah secara real-time yang sudah terbukti sangat efisien dalam klasifikasi wajah. Metode diimplementasikan dengan bantuan library OpenCV untuk deteksi multi wajah dan perangkat Web Cam M-Tech 5MP. Dalam penyusunan arsitekur model Convolutional Neural Networks dilakukan konfigurasi inisialisasi parameter untuk mempercepat proses training jaringan. Hasil uji coba dengan munggunakan konstruksi model Convolutional Neural Networks sampai kedalaman 7 lapisan dengan input dari hasil ekstraksi Extended Local Binary Pattern dengan radius 1 dan neighbor 15 menunjukkan kinerja pengenalan wajah meraih rata-rata tingkat akurasi lebih dari 89% dalam ∓ 2 frame per detik. ======================================================================================================= Identification of individual identity through facial recognition automatically is a huge deal of interest and a lot of the various approaches to resolving the problems. Moreover, in real life scenarios that are not controlled, the face will be visible from all sides and not always facing the front which makes it more difficult classification problems resolved. This final project use one of the methods of deep neural networks that Convolutional Neural Networks (CNN) as face recognition in real-time which has proven very efficient in the face classification. The method is implemented with the help of OpenCV library for multi face detection devices and a 5MP MTech Web Cam. In the preparation of architectural models Convolutional Neural Networks do initial configuration parameters to speed up the training process of the network. The trial results with munggunakan model construction Convolutional Neural Networks to a depth of 7 layers with input from the extraction Extended Local Binary Pattern with radius 1 and neighbor 15 shows the performance of face recognition reached average accuracy rate of more than 89% in ∓ 2 frames per second.

Item Type: Thesis (Undergraduate)
Additional Information: RSMa 006.33 Zuf c-1
Uncontrolled Keywords: Pengenalan Wajah, Real-Time, Convolutional Neural Networks
Subjects: Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions: Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User: Mr. Tondo Indra Nyata
Date Deposited: 06 Jan 2020 05:56
Last Modified: 06 Jan 2020 05:56
URI: http://repository.its.ac.id/id/eprint/72552

Actions (login required)

View Item View Item