Skripsi

PENGEMBANGAN ARSITEKTUR AC-GAN BERBASIS CONDITIONAL BATCH NORMALIZATION UNTUK AUGMENTASI CITRA PENYAKIT DAUN KELAPA SAWIT

---

The limited quantity and variety of oil palm leaf disease image data pose a problem for the training process, which requires large, diverse datasets. The development of generative methods that can increase the diversity and quality of synthetic image data is important. This study aims to develop an Auxiliary Classifier Generative Adversarial Network (AC-GAN) architecture based on Conditional Batch Normalization (CBN) as an augmentation method for oil palm leaf disease image data. The application of CBN to the generator is expected to improve the quality of the resulting synthetic images and maintain the consistency of class information. The generator in AC-GAN produces synthetic images through several learning layers, while the discriminator functions to distinguish between original and synthetic images and performs class classification through an auxiliary classifier. This method is applied to oil palm leaf image data grouped into four disease classes, namely dryness, fungal disease, magnesium deficiency, and scale insects. The success of the CBN-based AC-GAN architecture development is evaluated using the performance parameters Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Fréchet Inception Distance (FID). The results of the study show that the AC-GAN can produce high-quality synthetic images. The PSNR value of 34.8 dB indicates a low level of image formation error; the SSIM value of 0.89 indicates a high level of structural similarity between the synthetic and original images, and the FID value of 18.94 indicates that the distribution of synthetic image features is close to that of the original image. This shows that applying an AC-GAN is effective for augmenting oil palm leaf disease image data and has the potential to improve the quality of training data for learning models.

---

Availability

Detail Information

Series Title

-

Call Number

T1944062026

Publisher

Indralaya : Prodi Ilmu Matematika, Fakultas Matematika Dan Ilmu Pengetahuan Alam Universitas Sriwijaya., 2026

Collation

xviii, 178 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

006.370 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Computer Vision
Prodi Ilmu Matematika

Specific Detail Info

-

Statement of Responsibility

TUTI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment