PENERAPAN METODE CONVOLUTIONAL NEURAL NETWORK (CNN) DALAM PENGENALAN EMOSI WAJAH PADA PEMBELAJARAN DARING UNTUK IDENTIFIKASI TINGKAT KONSENTRASI PESERTA DIDIK

Aini, Isna Nur (2022) PENERAPAN METODE CONVOLUTIONAL NEURAL NETWORK (CNN) DALAM PENGENALAN EMOSI WAJAH PADA PEMBELAJARAN DARING UNTUK IDENTIFIKASI TINGKAT KONSENTRASI PESERTA DIDIK. Other thesis, UPN 'Veteran" Yogyakarta.

	Text COVER.pdf Download (121kB)
	Text DAFTAR PUSTAKA.pdf Download (145kB)
	Text PENGESAHAN PEMBIMBING.pdf Download (373kB)
	Text SKRIPSI FULL_ISNA NUR AINI.pdf Restricted to Repository staff only Download (14MB)
	Text ABSTRAK.pdf Download (15kB)
	Text DAFTAR ISI.pdf Download (99kB)

Abstract

vii
ABSTRAK
Pembelajaran daring menjadi solusi yang diterapkan dalam penyelenggaraan
kegiatan belajar mengajar guna menghindari penyebaran virus Covid-19. Selama
pembelajaran daring berlangsung, teridentifikasi beberapa permasalahan diantaranya
kosentrasi yang menurun dan hambatan dalam memberikan feedback verbal maupun non�verbal. Beberapa penelitian sebelumnya membuktikan adanya pengenalan emosi berdasar
ekspresi wajah dapat memberikan informasi kepada pengajar berupa feedback non-verbal
serta tingkat konsentrasi peserta didik. Dalam beberapa penelitian terkait pengenalan emosi,
terdapat permasalahan seperti overfitting dan permasalahan lain yang tidak terkait dengan
emosi, salah satunya variasi pose kepala.
Pada penelitian ini, metode Viola-Jones dan Convolutional Neural Network (CNN)
diterapkan dalam proses pengenalan emosi wajah melalui sebuah sistem monitoring emosi
dan tingkat konsentrasi pada pembelajaran daring. Penggunaan gabungan cascade classifier
diterapkan dalam sistem untuk dapat mendeteksi wajah dengan pose kepala beragam.
Beberapa teknik regularisasi diterapkan dalam pembuatan model untuk menghindarkan
model dari overfitting.
Hasil dari pengujian didapatkan bahwa metode yang diterapkan mampu mendeteksi
wajah dengan akurasi 98% pada sudut 0° dan 65% pada sudut >0°. Model yang dibuat dalam
penelitian ini mempunyai akurasi 67% dan penggunaan teknik regularisasi mampu
menghindarkan model dari overfitting. Implementasi sistem pada pembelajaran yang
dilakukan melalui Zoom Meeting mendapat hasil cukup baik, di mana dari 22 peserta yang
ada 18 wajah dapat terdeteksi dan dikenali emosinya.
Kata kunci : citra ekspresi wajah, hyperparameter, CNN, deteksi wajah, pengenalan
emosi wajah, Viola Jones
viii
ABSTRACT
Online learning is a solution that is applied in the implementation of teaching and
learning activities to avoid the spread of the Covid-19 virus. During online learning, several
problems were identified, including decreased concentration and obstacles in providing
verbal and non-verbal feedback. Several previous studies have proven that the recognition
of emotions based on facial expressions can provide information to teachers in the form of
non-verbal feedback and the level of concentration of students. In several studies related to
emotion recognition, there are problems such as overfitting and other problems that are not
related to emotions, one of which is variations in head poses.
In this study, the Viola-Jones method and the Convolutional Neural Network (CNN)
were applied in the facial emotion recognition process through an emotion monitoring
system and the level of concentration in online learning. The combined use of a cascade
classifier is implemented in the system to be able to detect faces with various head poses.
Several regularization techniques are applied in modeling to prevent the model from being
overfitted.
The results of the test showed that the method applied was able to detect faces with
an accuracy of 98% at an angle of 0° and 65% at an angle of >0°. The model made in this
study has an accuracy of 67% and the use of regularization techniques is able to prevent the
model from overfitting. The implementation of the system in learning carried out through
Zoom Meetings got quite good results, where from 22 participants there were 18 faces that
could be detected and their emotions recognized.
Keywords: facial expression image, hyperparameter, CNN, face detection, facial emotion
recognition, Viola Jones

Item Type:	Thesis (Other)
Uncontrolled Keywords:	facial expression image, hyperparameter, CNN, face detection, facial emotion recognition, Viola Jones
Subjects:	Z Bibliography. Library Science. Information Resources > ZA Information resources
Divisions:	Faculty of Engineering, Science and Mathematics > School of Engineering Sciences
Depositing User:	Eko Yuli
Date Deposited:	11 Nov 2022 03:04
Last Modified:	11 Nov 2022 03:04
URI:	http://eprints.upnyk.ac.id/id/eprint/31585

Actions (login required)

View Item