Yahya, Arif Bachtiar (2021) IMPLEMENTASI ALGORITMA DEEP CONVOLUTIONAL GENERATIVE ADVERSARIAL NETWORK (DCGAN) SEBAGAI GENERATOR GAMBAR BUNGA MAWAR. Undergraduate thesis, Universitas Stikubank.
![[thumbnail of HLM JUDUL]](/style/images/fileicons/application_pdf.png) |
PDF (HLM JUDUL)
Download (973kB) |
|
PDF (ABSTRAK)
Download (163kB) |
|
PDF (BAB I)
Download (118kB) |
|
PDF (BAB II)
Restricted to Repository staff only Download (89kB) |
|
PDF (BAB III)
Restricted to Repository staff only Download (347kB) |
|
PDF (BAB IV)
Restricted to Repository staff only Download (207kB) |
|
PDF (BAB V)
Restricted to Repository staff only Download (344kB) |
|
PDF (BAB VI)
Restricted to Repository staff only Download (8kB) |
|
PDF (DAFTAR PUSTAKA)
Download (77kB) |
|
PDF (LAMPIRAN)
Restricted to Repository staff only Download (9MB) |
Abstract
Penelitian terhadap Deep learning banyak dilakukan untuk mendukung perkembangan Artificial Intelligence, namun beberapa penelitian mengalami kendala karena kekurangan dataset untuk bisa dipelajari dan keterbatasan akses ke dalam pustaka sumber dataset. Salah satu cara untuk mengatasi hal tersebut yaitu dengan menciptakan metode lain yang dapat menghasilkan gambar baru berdasarkan dataset yang sudah ada bernama Generative Adversarial Network (GAN). Namun model GAN ini masih memiliki kelemahan yaitu kurang stabilnya model sehinga menyebabkan hasil gambar yang tidak jelas. Penelitian ini menggunakan algoritma Deep Convolutional Generative Adversarial Network (DCGAN) yang merupakan gabungan teknik konvolusi dari Convolutional Neural Network(CNN) dan model Generative Adversarial Network (GAN) dalam menghasilkan gambar baru. Model generator akan memanfaatkan teknik transpos konvolusi dalam menghasilkan gambar sedangkan model discriminator akan menggunakan teknik konvolusi downsampling untuk meninjau gambar yang dihasilkan model generator. Penelitian ini menggunakan dataset gambar bunga mawar sebagai objek, dan terbukti bahwa algoritma DCGAN dapat digunakan untuk menghasilkan gambar baru bunga mawar sesuai dataset yang disediakan. A lot of research on deep learning has been carried out to support the development of Artificial Intelligence, but some researches have had problems due to the lack of datasets to study and limited access to the dataset source library. One way to overcome this is to create another method that can generate new images based on an existing dataset called the Generative Adversarial Network (GAN). However, this GAN model still has a weakness, namely the lack of stability of the model so that it causes unclear image results. This study uses the Deep Convolutional Generative Adversarial Network (DCGAN) algorithm which is a combination of convolutional techniques from Convolutional Neural Network (CNN) and Generative Adversarial Network (GAN) models in generating new images. The generator model will utilize the convolution transpose technique in producing the image while the discriminator model will use the downsampling convolution technique to review the resulting generator model image. This study uses a rose image dataset as an object, and it is proven that the DCGAN algorithm can be used to generate new rose images according to the provided dataset.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | SKR.I.05.01.2037 NIM 17.01.53.0089 |
| Uncontrolled Keywords: | Generative Adversarial Network, Deep Learning, Artificial Intelligence, Deep Convolutional Generative Adversarial Network, Gambar Baru Generative Adversarial Network, Deep Learning, Artificial Intelligence, Deep Convolutional Generative Adversarial Network, New Images |
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science |
| Faculty / Institution: | Fakultas Teknologi Informasi > Program Studi Teknik Informatika |
| Depositing User: | Teteh Hayati |
| Date Deposited: | 10 Nov 2021 07:11 |
| Last Modified: | 10 Nov 2021 07:11 |
| URI: | https://eprints.unisbank.ac.id/id/eprint/8038 |
Actions (login required)
 |
View Item |
