<![CDATA[Cocoa plant (Theobroma Cacao) is an important economic commodity in indonesia that is susceprible to pest and disease attacks, particularly the Cocoa Pod Borer (CPB) which can reduce crop yileds by up to 70%. Limited access to agricultural experts and insufficient datasets pose challenges in applying deep learning technology for plant disease diagnosis. This research aims to implment Generative Adversarial Networks (GANs) method, specifically Deep Convolutional GAN (DCGAN) , for sataset augmentation to improve the performance of an expert system for diagnosing cocoa plant diseases. The research methodology includes data collection through field observations in Atar Lebar Village, Lampung and public datasets from Kaggle, image preprcessing (resizing to 128 x 128 pixels and normalization from -1 to 1 ), augmnetation using DCGAN to generate 2,000 synthetic images for disease class, and clasification using Convolutional Neural Network CNN) and MobileNetV2. The system is implmented as aweb -based application using PHP and Python for diagnosing three main disease types: Pod Borer, Monilia and Black Pod Rot, along wirh Healthy and Non – Cocoa Classes.]]>
Copyrights © 2026
