Pengenalan Ekspresi Wajah Menggunakan Wavelet Transform Dan Convolutional Neural Network

Wiranto, Hendry (2019) Pengenalan Ekspresi Wajah Menggunakan Wavelet Transform Dan Convolutional Neural Network. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 05111540000102-Undergraduate_Theses.pdf]
Preview
Text
05111540000102-Undergraduate_Theses.pdf

Download (5MB) | Preview

Abstract

Machine learning telah menjadi bagian dari kehidupan sehari-hari bagi banyak orang. Salah satu pengaplikasian machine learning adalah pengenalan ekspresi wajah manusia. Pengenalan ekspresi wajah manusia mengkategorikan gambar ekspresi wajah menjadi satu dari banyak kelas ekspresi wajah berdasarkan fitur gambar tersebut. Banyak perusahaan, badan riset dan universitas yang terus mengembangkan machine learning agar mendapat hasil yang lebih akurat dan cepat. Dari situlah lahir algoritma deep learning, yang merupakan bagian dari machine learning. Convolutional Neural Network (CNN) adalah salah satu deep neural network yang cocok digunakan untuk mengolah data yang berbentuk 2 dimensi, seperti gambar dan video.
Pada tugas akhir ini, penulis mengusulkan sebuah algoritma untuk mengubah data gambar menjadi Wavelet Domain dengan menggunakan Wavelet Transform. Tujuannya adalah untuk meningkatkan akurasi pengenalan ekspresi wajah manusia dengan metode Convolutional Neural Network. Data pelatihan dan uji coba diambil dari dataset “Karolinska Directed Emotional Faces” (KDEF) yang berisi foto wajah manusia dengan 7 ekspresi berbeda yang nantinya menjadi tujuan pengenalan ekspresi wajah manusia yang dibuat. Praproses terhadap data antara lain dilakukan perubahan format gambar menjadi grayscale, perubahan resolusi gambar menjadi 256x256 piksel, dilakukan proses Discrete Wavelet Transform level 1, dan dilakukan proses augmentasi data berupa refleksi horizontal dan perbesaran ukuran gambar. Hasil uji coba terakhir didapatkan nilai akurasi 89,6%.
================================================================================================
Machine learning has become a part of the daily life of people around the world. One of the application of machine learning is human facial expression recognition. Human facial expression recognition categorizes an image of facial expression into one of many facial expression classes based on the features extracted from the image. Many companies, researchers and universities keep improving the machine learning to get a better and faster result. And from those improvements, deep learning algorithm is born. Convolutional Neural Network (CNN) is one of the deep neural network that suitable to process 2 dimentional data like image and video.
In this undergraduate thesis, the images are transformed into Wavelet Domain using Wavelet Transform before being processed into the proposed network. The purpose of this method is to improve the accuracy of the human facial expression recognition using Convolutional Neural Network. The train and test data used in this thesis is taken from “Karolinska Directed Emotional Faces” (KDEF) dataset which contains human facial expression with 7 different expressions which will be the prediction labels of the human facial expression recognition. The preprocessing of the images include changing the image format to grayscale, changing the image resolution to 256x256 pixels, applying level 1 Discrete Wavelet Transform and applying data augmentation with horizontal reflection and zoom in. The final test accuracy is 89,6%.

Item Type: Thesis (Undergraduate)
Additional Information: RSIf 006.33 Wir p-1 2019
Uncontrolled Keywords: Convolutional Neural Network, Data Gambar, Dataset Karolinska Directed Emotional Faces, Pengenalan ekspresi wajah manusia, Wavelet Transform
Subjects: Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
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.
Divisions: Faculty of Information and Communication Technology > Informatics > 55201-(S1) Undergraduate Thesis
Depositing User: Hendry Wiranto
Date Deposited: 09 Jul 2021 04:29
Last Modified: 09 Jul 2021 04:29
URI: http://repository.its.ac.id/id/eprint/60805

Actions (login required)

View Item View Item