i




SKRIPSI

Diajukan Sebagai Salah Satu Syarat untuk Memperoleh Gelar

Sarjana Komputer

MARSHA NADIRA

1510511031

UNIVERSITAS PEMBANGUNAN NASIONAL “VETERAN” JAKARTA

FAKULTAS ILMU KOMPUTER

PROGRAM STUDI INFORMATIKA

2019

ii

PERNYATAAN ORISINALITAS

iii

PERNYATAAN

PERSETUJUAN PUBLIKASI

SKRIPSI UNTUK

KEPENTINGAN AKADEMIS

iv

PENGESAHAN

v




MARSHA NADIRA

ABSTRAK

Di Indonesia, kulit hewan merupakan salah satu bahan mentah yang digunakan sebagai

bahan utama dalam industri kulit dan karya seni. Selama proses pengolahan kulit

hewan menjadi kulit jadi asli, kulit hewan tersebut telah mengalami perubahan baik

tekstur maupun warna. Hal itu menyebabkan sulit dikenalinya jenis kulit hewan antar

satu spesies dengan spesies lainnya. Untuk mengatasi permasalahan tersebut,

dibutuhkan suatu sistem yang dapat mengidentifikasi jenis bahan kulit hewan.

Perkembangan teknologi dan ilmu pengetahuan yang pesat memungkinkan untuk dapat

melakukan klasifikasi suatu objek berdasarkan karakteristik yang dimilikinya. Salah

satu metode deep learning yang sedang berkembang saat ini adalah Convolutional

Neural Network (CNN). CNN diklaim merupakan metode terbaik dalam permasalahan

deteksi dan pengenalan objek. Dengan menggunakan citra sebagai masukannya, CNN

dapat melakukan pembelajaran citra sehingga dapat menghasilkan sebuah model yang

mampu mengklasifikasikan objek citra jenis bahan kulit hewan. Konvolusi merupakan

proses utama yang mendasari jaringan arsitektur CNN, sehingga citra dapat diekstrasi

setiap fiturnya dengan lebih baik dan mempermudah proses klasifikasi. Penelitian ini

diharapkan dapat membuat sistem yang mampu mengenali jenis bahan kulit hewan

dengan tepat dan memiliki akurasi yang tinggi.

Kata Kunci: pengolahan citra, convolutional neural network (CNN), deep learning,

kulit hewan

vi

DEEP LEARNING IMPLEMENTATION WITH

CONVOLUTIONAL NEURAL NETWORK METHOD FOR

LEATHER IMAGE IDENTIFICATION

MARSHA NADIRA

ABSTRACT

In Indonesia, animal skin is one of raw material which is used as main material in

leather and art industry. During the process of processing animal skin into genuine

leather, the animal skin has undergone changes in both texture and color. It causes

difficulties in identifiying leather on each species. To overcome that problem, is needed

a system to identify leather types. Rapid development of technology and science make

it possible to classify object based on their characteristics. One of the deep learning

methods which is currently developing is Convolutional Neural Network (CNN). CNN

is claimed as the best method for detection and object recognition problem. Using

image as an input, CNN can do image learning to generate a model to classify leather

images. Convolution is the main process underlying CNN network architecture, so that

each of images can be extracted better to simplify the classification process. This

research is expected to create a system which can identify leather types appropriately

and has high accuracy.

Keywords: image processing, convolutional neural network (CNN), deep learning,

leather

vii

KATA PENGANTAR

Puji dan syukur penulis panjatkan kehadirat Allah SWT atas segala karunia-Nya,

sehingga Skripsi ini berhasil diselesaikan. Penulis ingin mengucapkan terima kasih

kepada:

1. Orang tua dan keluarga yang selalu memberikan dorongan kepada penulis agar

dapat menyelesaikan skripsinya.

2. Ibu Vini Indriasari, S.T., M.Sc., Ph.D., selaku dosen pembimbing skripsi yang

telah memberikan saran yang bermanfaat.

3. Ibu Dr. Ermatita, M. Kom., selaku Dekan Fakultas Ilmu Komputer.

4. Ibu Anita Muliawati, S.Kom., MTI., selaku Ketua Program Studi S1

Informatika.

5. Ibu Mayanda Mega Santoni, S.Kom., M.Kom., selaku dosen Fakultas Ilmu

Komputer yang bersedia menjadi rekan diskusi.

6. Teman-teman Program Studi S1 Informatika Angkatan 2015 Fakultas Ilmu

Komputer yang telah memberikan dukungan, semangat, dan doa.

7. Teman-teman BEMF-IK periode 2017-2018 dan UKM Fotografi & Videografi

“Veteran” Jakarta yang telah memberikan dukungan, semangat, dan doa.

Dan semua pihak yang telah membantu penulis dalam menyelesaikan Skripsi ini.

Jakarta, 16 Juli 2019

Penulis

viii

DAFTAR ISI

HALAMAN SAMPUL .................................................................................................. i

PERNYATAAN ORISINALITAS ............................................................................... ii

PERNYATAAN PERSETUJUAN PUBLIKASI SKRIPSI UNTUK KEPENTINGAN

AKADEMIS ................................................................................................................ iii

PENGESAHAN ........................................................................................................... iv

ABSTRAK .................................................................................................................... v

ABSTRACT ................................................................................................................. vi

KATA PENGANTAR ................................................................................................ vii

DAFTAR ISI .............................................................................................................. viii

DAFTAR TABEL ....................................................................................................... xii

DAFTAR GAMBAR ................................................................................................. xiii

BAB 1 PENDAHULUAN ............................................................................................ 1

1.1 Latar Belakang Masalah.................................................................................... 1

1.2 Rumusan Masalah ............................................................................................. 3

1.3 Tujuan Penelitian .............................................................................................. 3

1.4 Manfaat Penelitian ............................................................................................ 3

1.5 Ruang Lingkup .................................................................................................. 4

1.6 Luaran yang Diharapkan ................................................................................... 4

1.7 Sistematika Penulisan ....................................................................................... 4

BAB 2 TINJAUAN PUSTAKA ................................................................................... 6

2.1 Citra Digital ...................................................................................................... 6

2.1.1 Citra Biner ............................................................................................... 7

ix

2.1.2 Citra Grayscale ....................................................................................... 8

2.1.3 Citra Warna RGB .................................................................................... 9

2.1.4 Citra Warna Berindeks ............................................................................ 9

2.2 Pengolahan Citra ............................................................................................. 10

2.3 Klasifikasi Citra Menggunakan Convolutional Neural Network .................... 12

2.3.1 Operasi Konvolusi ................................................................................ 14

2.3.2 Operasi Pooling .................................................................................... 16

2.3.3 Aktivasi ReLU ...................................................................................... 17

2.3.4 Fully-Connected Layer ......................................................................... 18

2.3.5 Softmax Classifier ................................................................................. 18

2.3.6 Mean Squared Error (MSE) ................................................................. 19

2.3.7 Confusion Matrix .................................................................................. 19

2.3.8 Akurasi .................................................................................................. 20

2.4 Library untuk Menyelesaikan Masalah Jaringan Syaraf: Keras ..................... 20

2.5 Metode Klasifikasi Citra untuk Identifikasi Kulit .......................................... 21

2.6 Penerapan Metode CNN untuk Klasifikasi Citra ............................................ 22

BAB 3 METODOLOGI PENELITIAN ..................................................................... 25

3.1 Kerangka Pikir ................................................................................................ 25

3.1.1 Identifikasi dan Perumusan Masalah .................................................... 26

3.1.2 Studi Pustaka ......................................................................................... 26

3.1.3 Pengumpulan Data ................................................................................ 26

3.1.4 Pengolahan Data ................................................................................... 26

3.1.5 Perancangan CNN ................................................................................. 27

3.1.6 Pelatihan dan Pengujian Model ............................................................ 27

x

3.1.7 Perhitungan Akurasi .............................................................................. 27

3.1.8 Analisis Hasil ........................................................................................ 28

3.1.9 Kesimpulan dan Saran .......................................................................... 28

3.2 Perangkat Penelitian ........................................................................................ 28

3.2.1 Perangkat Keras (Hardware) ................................................................ 28

3.2.2 Perangkat Lunak (Software) ................................................................. 29

BAB 4 HASIL DAN PEMBAHASAN ...................................................................... 30

4.1 Pengumpulan Data Citra ................................................................................. 30

4.2 Pengolahan Data Citra .................................................................................... 32

4.3 Perancangan CNN ........................................................................................... 33

4.4 Pembagian Data Latih dan Data Uji ............................................................... 34

4.5 Pelatihan Model .............................................................................................. 37

4.6 Pengujian Model ............................................................................................. 44

4.7 Perhitungan Akurasi ........................................................................................ 49

4.8 Penentuan Parameter Model ........................................................................... 50

4.8.1 Pengaruh Ukuran Batch Size ................................................................. 51

4.8.2 Pengaruh Jumlah Epoch ........................................................................ 52

4.8.3 Pengaruh Nilai Learning Rate............................................................... 53

4.8.4 Pengaruh Pembagian Data Latih dan Data Uji ..................................... 54

BAB 5 PENUTUP ...................................................................................................... 55

5.1 Kesimpulan ..................................................................................................... 55

5.2 Saran ............................................................................................................... 55

DAFTAR PUSTAKA ................................................................................................. 57

RIWAYAT HIDUP .................................................................................................... 60

xi

LAMPIRAN ................................................................................................................ 62

xii

DAFTAR TABEL

Tabel 2.1 Model Confusion Matrix ............................................................................. 20

Tabel 2.2 Perbandingan Studi Relevan ....................................................................... 24

Tabel 4.1 Perbandingan Data Latih dan Data Uji ....................................................... 36

Tabel 4.2 Hasil Akurasi Data Latih dan Validasi ....................................................... 44

Tabel 4.3 Confusion Matrix Seluruh Kelas ................................................................. 47

Tabel 4.4 Confusion Matrix Kelas Kulit Sapi ............................................................. 48

Tabel 4.5 Confusion Matrix Kelas Kulit Babi ............................................................ 48

Tabel 4.6 Confusion Matrix Kelas Kulit Domba ........................................................ 48

Tabel 4.7 Confusion Matrix Kelas Kulit Kambing ..................................................... 49

Tabel 4.8 Confusion Matrix Kelas Kulit Kanguru ...................................................... 49

Tabel 4.9 Akurasi Berdasarkan Ukuran Batch Size .................................................... 51

Tabel 4.10 Akurasi Berdasarkan Jumlah Epoch ......................................................... 52

Tabel 4.11 Akurasi Berdasarkan Nilai Learning Rate ................................................ 53

Tabel 4.12 Akurasi Berdasarkan Pembagian Data Latih dan Data Uji ....................... 54

xiii

DAFTAR GAMBAR

Gambar 2.1 Representasi Citra Digital dalam Matriks ................................................. 7

Gambar 2.2 Citra Biner ................................................................................................. 8

Gambar 2.3 Citra Grayscale ......................................................................................... 8

Gambar 2.4 Citra Warna RGB ...................................................................................... 9

Gambar 2.5 Citra Warna Berindeks ............................................................................ 10

Gambar 2.6 Arsitektur Convolutional Neural Network .............................................. 13

Gambar 2.7 Operasi pada Convolution Layer ............................................................. 16

Gambar 2.8 Operasi pada Max Pooling ...................................................................... 17

Gambar 3.1 Kerangka Pikir ........................................................................................ 25

Gambar 4.1 Citra Bahan Kulit Sapi ............................................................................ 31

Gambar 4.2 Citra Bahan Kulit Babi ............................................................................ 31

Gambar 4.3 Citra Bahan Kulit Domba ....................................................................... 31

Gambar 4.4 Citra Bahan Kulit Kambing .................................................................... 31

Gambar 4.5 Citra Bahan Kulit Kanguru ..................................................................... 32

Gambar 4.6 Contoh Hasil Cropping Citra Bahan Kulit Hewan ................................. 32

Gambar 4.7 Flowchart Model CNN ........................................................................... 33

Gambar 4.8 Tabel Data Citra ...................................................................................... 35

Gambar 4.9 Hasil Train Test Split Data ...................................................................... 36

Gambar 4.10 Summary Model CNN ........................................................................... 42

Gambar 4.11 Nilai Akurasi dan Loss pada Model ...................................................... 43

convolutional neural network untuk identifikasi …

Documents