ANALISIS CITRA PADA POTRET BOTOL MINUMAN BEKAS MENGGUNAKAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK

RAMADHANI, NANDIKO (2023) ANALISIS CITRA PADA POTRET BOTOL MINUMAN BEKAS MENGGUNAKAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK. S1 thesis, Universitas Mercu Buana Jakarta.

[img]
Preview
Text (HAL COVER)
01 Cover.pdf

Download (378kB) | Preview
[img]
Preview
Text (ABSTRAK)
02 Abstrak.pdf

Download (119kB) | Preview
[img] Text (BAB I)
03 Bab 1.pdf
Restricted to Registered users only

Download (187kB)
[img] Text (BAB II)
04 Bab 2.pdf
Restricted to Registered users only

Download (858kB)
[img] Text (BAB III)
05 Bab 3.pdf
Restricted to Registered users only

Download (852kB)
[img] Text (BAB IV)
06 Bab 4.pdf
Restricted to Registered users only

Download (513kB)
[img] Text (BAB V)
07 Bab 5.pdf
Restricted to Registered users only

Download (116kB)
[img] Text (DAFTAR PUSTAKA)
08 Daftar Pustaka.pdf
Restricted to Registered users only

Download (191kB)
[img] Text (LAMPIRAN)
09 Lampiran.pdf
Restricted to Registered users only

Download (251kB)

Abstract

The presence of human life that uses various types of products has the potential to produce multiple types of waste. The official website of the National Waste Management Information System (SIPSN) explains that the second place in national waste production in 2021 is plastic waste, with a percentage of 17.47%. Plastic-based beverage bottle waste takes a long time to decompose. One of the solutions to overcome this is by recycling waste. Some recycling managers still use manual waste sorting techniques. Therefore, a classification system is needed to make sorting the used beverage bottle waste easier. In this research, the system design was carried out using the CNN algorithm, in which there are several single CNN models with the base model from MobileNetV2 and modified in the head model or fully-connected layer, resulting in the MobileNetV2 architecture with three types of head models, namely Head Model 1, Head Model 2, and Head Model 3. This research also proposes an ensemble method applied to all single CNN models by taking the weight values for each model obtained after training and processing the average weights to improve image classification performance. This method forms a new model known as the Ensemble Convolutional Neural Network (E-CNN) model. Based on the results of tests conducted in this research, the CNN model can classify images on portraits of used beverage bottles accurately and effectively. This is shown in the test classification report values, namely the use of a single CNN model in the form of the MobileNetV2 architecture with Head Model 1, Head Model 2, and Head Model 3, which have successive accuracy values of 91%, 89%, and 91%. Besides that, after applying the ensemble method and becoming an ECNN model, the test accuracy value is 98%, where the accuracy value increases by 7% to 9%. Keywords: Image classification, Convolutional Neural Network (CNN), MobileNetV2, ensemble method, bottled waste Hadirnya kehidupan manusia yang menggunakan berbagai jenis produk berpotensi menghasilkan beraneka ragam sampah. Dalam situs resmi Sistem Informasi Pengelolaan Sampah Nasional (SIPSN) menerangkan bahwa urutan kedua pada produksi sampah nasional tahun 2021 ialah sampah plastik dengan persentase sebesar 17.47%. Sampah botol minuman bekas berbahan dasar plastik membutuhkan waktu yang lama agar dapat terurai. Di antara solusi untuk mengatasi hal itu ialah dengan melakukan daur ulang sampah. Sebagian pengelola daur ulang masih menggunakan teknik penyortiran sampah secara manual. Oleh sebab itu, dibutuhkannya suatu sistem klasifikasi agar mempermudah dalam penyortiran sampah botol minuman bekas tersebut. Pada penelitian ini, dilakukan perancangan sistem dengan algoritma CNN, di mana terdapat beberapa model CNN tunggal dengan base model dari MobileNetV2 dan dimodifikasi pada bagian head model atau fully-connected layer, sehingga menghasilkan arsitektur MobileNetV2 dengan tiga macam head model, yaitu Head Model 1, Head Model 2, dan Head Model 3. Pada penelitian ini, juga diusulkan metode ensemble yang diterapkan pada seluruh model CNN tunggal dengan mengambil nilai bobot (weight) tiap model yang diperoleh setelah pelatihan, kemudian dilakukan proses bobot rata-rata (average weights) untuk meningkatkan performa pengklasifikasian gambar. Dari metode tersebut, maka terbentuk model baru yang dikenal dengan model Ensemble Convolutional Neural Network (ECNN). Berdasarkan hasil pengujian yang dilakukan pada penelitian ini, model CNN mampu mengklasifikasikan citra pada potret botol minuman bekas secara akurat dan efektif. Hal ini ditunjukkan pada nilai classification report pengujian, yaitu penggunaan model CNN tunggal berupa arsitektur MobileNetV2 dengan Head Model 1, Head Model 2, dan Head Model 3 yang memiliki nilai accuracy berturutturut sebesar 91%, 89%, dan 91%. Selain itu, setelah diterapkan metode ensemble dan menjadi suatu model E-CNN, maka didapatkan nilai accuracy pengujian sebesar 98%, di mana terjadi peningkatan nilai accuracy sebesar 7% hingga 9%. Kata kunci: Klasifikasi citra, Convolutional Neural Network (CNN), MobileNetV2, ensemble method, botol minuman bekas

Item Type: Thesis (S1)
Call Number CD: FT/ELK. 23 036
Call Number: ST/14/23/013
NIM/NIDN Creators: 41418010025
Uncontrolled Keywords: Klasifikasi citra, Convolutional Neural Network (CNN), MobileNetV2, ensemble method, botol minuman bekas
Subjects: 100 Philosophy and Psychology/Filsafat dan Psikologi > 150 Psychology/Psikologi > 154 Subconscious and Altered States and Process/Psikologi Bawah Sadar > 154.6 Sleep Phenomena/Fenomena Tidur > 154.63 Dreams/Mimpi > 154.634 Analysis/Analisis
500 Natural Science and Mathematics/Ilmu-ilmu Alam dan Matematika > 510 Mathematics/Matematika > 518 Numerical Analysis/Analisis Numerik, Analisa Numerik
500 Natural Science and Mathematics/Ilmu-ilmu Alam dan Matematika > 510 Mathematics/Matematika > 518 Numerical Analysis/Analisis Numerik, Analisa Numerik > 518.1 Algorithms/Algoritma
600 Technology/Teknologi > 620 Engineering and Applied Operations/Ilmu Teknik dan operasi Terapan
600 Technology/Teknologi > 620 Engineering and Applied Operations/Ilmu Teknik dan operasi Terapan > 621 Applied Physics/Fisika terapan
600 Technology/Teknologi > 620 Engineering and Applied Operations/Ilmu Teknik dan operasi Terapan > 621 Applied Physics/Fisika terapan > 621.3 Electrical Engineering, Lighting, Superconductivity, Magnetic Engineering, Applied Optics, Paraphotic Technology, Electronics Communications Engineering, Computers/Teknik Elektro, Pencahayaan, Superkonduktivitas, Teknik Magnetik, Optik Terapan, Tekn
Divisions: Fakultas Teknik > Teknik Elektro
Depositing User: ADELINA HASNA SETIAWATI
Date Deposited: 08 Mar 2023 02:54
Last Modified: 16 Mar 2023 07:57
URI: http://repository.mercubuana.ac.id/id/eprint/74840

Actions (login required)

View Item View Item