Implementasi Deep Learning Berbasis Keras Untuk Pengenalan Wajah

Santoso, Aditya and -, Gunawan Ariyanto,ST. M. Compt Sc., Ph.D. (2018) Implementasi Deep Learning Berbasis Keras Untuk Pengenalan Wajah. Skripsi thesis, Universitas Muhammadiyah Surakarta.

[img] PDF (Naskah Publikasi)

Download (1MB)
[img] PDF (Surat Pernyataan Naskah Publikasi)
Surat pernyataan publikasi ilmiah.pdf
Restricted to Repository staff only

Download (17kB) | Request a copy


Facial recognition system is an important aspect in the field of computer vision that supports the development of sophisticated technology such as the current era. The use of face because the face has a uniqueness and become an identity for every human being. In its development, facial recognition systems still have problems in lighting factors, facial expression and attributes on the face. In this study the author uses Conventional Neural Network (CNN) to do this. CNN is part of the in-depth learning that is used to perform the learning process on the computer to find the best reprentation. CNN consists of 3 stages, namely Input data, Learning Features, and Classification. Each input data will go through the same process as the filtering process. Implementation of CNN in this study using a library that uses python programming language. Hard is a framework created to facilitate learning on computer. The dataset used in this study was face94 by taking 10 male face subjects. The CNN training process uses a 28x28 px data size with 7 layers which yields better measurements using 5 layers with 8.0% yield difference at the time of testing. The use of 7 layers during testing of good test results data with accuracy rate of 98.57%.

Item Type: Karya ilmiah (Skripsi)
Uncontrolled Keywords: Sistem pengenalan wajah, Convolutional Neural Network, Keras, Python.
Subjects: T Technology > T Technology (General)
Divisions: Fakultas Ilmu Komunikasi dan Informatika > Teknik Informatika
Depositing User: ADITYA SANTOSO
Date Deposited: 11 Jul 2018 03:49
Last Modified: 11 Jul 2018 03:49

Actions (login required)

View Item View Item