Abstract:
This project aims to develop an efficient and cost-effective system for the early detection and monitoring of oil palm diseases (Anthracnose, Bagworm, Curvularia leaf spot and Pestalotiopsis), which are major contributors to the decline in agriculture production. The proposed method combines computer vision techniques, specifically Convolutional Neural Networks (CNN), with a custom-built drone equipped with obstacle avoidance and autopilot features for disease detection in real-world conditions. The system combines image classification (VGG16 and AlexNet) and object detection (YOLOv3) to identify and locate the type of oil palm disease accurately and precisely, enabling farmers to take necessary actions to prevent the spread of diseases and ensure better yield. The study assesses the performance of various CNN architectures for image classification by analyzing the impact of different parameters, including the number of parameters, epochs, batch size, and learning rate. The study's potential impact is significant in the Malaysian economy's palm oil sector, contributing to the country's economic growth. Additionally, the study can provide valuable insights for future advancements in drone technology, particularly within the scope of Industry 4.0. The expected outcome of this project is to contribute to developing a more efficient and effective system for the early detection and monitoring of oil palm diseases, ultimately leading to increased productivity and profitability of the palm oil industry.
