Development of a mobile application for disease detection in oregano leaves using ylov8 model, tensorflow, and opencv library

ABSTRACT: Oregano, valued for its culinary and medicinal properties, is susceptible to stress-induced diseases that affect its health and yield. Detecting these diseases is challenging due to the variations in image quality and the need for accurate classification. To address this, advancements in image processing and machine learning provide a promising solution for improving disease detection for Oregano plants. This study explores the development of a mobile application that detects oregano leaf diseases using the YOLOv8 model for accurate classification. It also examines how image quality can be enhanced using OpenCV preprocessing technique to improve detection accuracy. Additionally, the study investigates how the application can provide disease information, including causes, symptoms, and recommended treatments, using TensorFlow. By integrating these technologies, the application aims to improve plant healthy through proper diagnosis and timely intervention. To evaluate the effectiveness of this approach, the study assessed the performance of the model and image enhancement techniques. The YOLOv8 model achieved 90.48% accuracy, with a precision of 80.61%, recall of 76%, and F1 score of 86.30%, highlighting reliable disease identification despite some false negatives. Image enhancement techniques using OpenCV were evaluated using MSE, PSNR, and SSIM, yielding 0.26, 27.13, and 0.93, respectively, indicating effective image quality preservation. Although limitations in handling false negatives and maintaining image clarity under challenging conditions were noted, the findings demonstrate the potential of the proposed application to enhance plant disease detection and support sustainable agriculture.

Filipiniana